CPACR_EL1 has several bitfields for controlling traps for floating point features to EL1, each of which has a separate bits for EL0 and EL1. Marc Zyngier noted that we are not consistent in our use of defines to manipulate these, sometimes using a define covering the whole field and sometimes using defines for the individual bits. Make this consistent by expanding the whole field defines where they are used (currently only in the KVM code) and deleting them so that no further uses can be introduced.

Suggested-by: Marc Zyngier maz@kernel.org Signed-off-by: Mark Brown broonie@kernel.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/kvm_arm.h | 3 ++- arch/arm64/include/asm/sysreg.h | 2 -- arch/arm64/kvm/hyp/include/hyp/switch.h | 4 ++-- arch/arm64/kvm/hyp/vhe/switch.c | 6 +++--- 4 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/arch/arm64/include/asm/kvm_arm.h b/arch/arm64/include/asm/kvm_arm.h index eec790842fe2..1767ded83888 100644 --- a/arch/arm64/include/asm/kvm_arm.h +++ b/arch/arm64/include/asm/kvm_arm.h @@ -356,6 +356,7 @@ ECN(BKPT32), ECN(VECTOR32), ECN(BRK64)

#define CPACR_EL1_TTA (1 << 28) -#define CPACR_EL1_DEFAULT (CPACR_EL1_FPEN | CPACR_EL1_ZEN_EL1EN) +#define CPACR_EL1_DEFAULT (CPACR_EL1_FPEN_EL0EN | CPACR_EL1_FPEN_EL1EN |\ + CPACR_EL1_ZEN_EL1EN)

#endif /* __ARM64_KVM_ARM_H__ */ diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h index 1da4c43d597d..e66dd9ebc337 100644 --- a/arch/arm64/include/asm/sysreg.h +++ b/arch/arm64/include/asm/sysreg.h @@ -1099,11 +1099,9 @@

#define CPACR_EL1_FPEN_EL1EN (BIT(20)) /* enable EL1 access */ #define CPACR_EL1_FPEN_EL0EN (BIT(21)) /* enable EL0 access, if EL1EN set */ -#define CPACR_EL1_FPEN (CPACR_EL1_FPEN_EL1EN | CPACR_EL1_FPEN_EL0EN)

#define CPACR_EL1_ZEN_EL1EN (BIT(16)) /* enable EL1 access */ #define CPACR_EL1_ZEN_EL0EN (BIT(17)) /* enable EL0 access, if EL1EN set */ -#define CPACR_EL1_ZEN (CPACR_EL1_ZEN_EL1EN | CPACR_EL1_ZEN_EL0EN)

/* TCR EL1 Bit Definitions */ #define SYS_TCR_EL1_TCMA1 (BIT(58)) diff --git a/arch/arm64/kvm/hyp/include/hyp/switch.h b/arch/arm64/kvm/hyp/include/hyp/switch.h index 701cfb964905..6379a1e3e6e5 100644 --- a/arch/arm64/kvm/hyp/include/hyp/switch.h +++ b/arch/arm64/kvm/hyp/include/hyp/switch.h @@ -174,9 +174,9 @@ static bool kvm_hyp_handle_fpsimd(struct kvm_vcpu *vcpu, u64 *exit_code)

/* Valid trap. Switch the context: */ if (has_vhe()) { - reg = CPACR_EL1_FPEN; + reg = CPACR_EL1_FPEN_EL0EN | CPACR_EL1_FPEN_EL1EN; if (sve_guest) - reg |= CPACR_EL1_ZEN; + reg |= CPACR_EL1_ZEN_EL0EN | CPACR_EL1_ZEN_EL1EN;

sysreg_clear_set(cpacr_el1, 0, reg); } else { diff --git a/arch/arm64/kvm/hyp/vhe/switch.c b/arch/arm64/kvm/hyp/vhe/switch.c index 11d053fdd604..619353b06e38 100644 --- a/arch/arm64/kvm/hyp/vhe/switch.c +++ b/arch/arm64/kvm/hyp/vhe/switch.c @@ -38,7 +38,7 @@ static void __activate_traps(struct kvm_vcpu *vcpu)

val = read_sysreg(cpacr_el1); val |= CPACR_EL1_TTA; - val &= ~CPACR_EL1_ZEN; + val &= ~(CPACR_EL1_ZEN_EL0EN | CPACR_EL1_ZEN_EL1EN);

/* * With VHE (HCR.E2H == 1), accesses to CPACR_EL1 are routed to @@ -53,9 +53,9 @@ static void __activate_traps(struct kvm_vcpu *vcpu)

if (update_fp_enabled(vcpu)) { if (vcpu_has_sve(vcpu)) - val |= CPACR_EL1_ZEN; + val |= CPACR_EL1_ZEN_EL0EN | CPACR_EL1_ZEN_EL1EN; } else { - val &= ~CPACR_EL1_FPEN; + val &= ~(CPACR_EL1_FPEN_EL0EN | CPACR_EL1_FPEN_EL1EN); __activate_traps_fpsimd32(vcpu); }

An ARRAY_SIZE() has been added to kselftest.h so remove the local versions in some of the arm64 selftests.

Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com Signed-off-by: Mark Brown broonie@kernel.org --- tools/testing/selftests/arm64/abi/syscall-abi.c | 1 - tools/testing/selftests/arm64/fp/sve-ptrace.c | 2 -- 2 files changed, 3 deletions(-)

diff --git a/tools/testing/selftests/arm64/abi/syscall-abi.c b/tools/testing/selftests/arm64/abi/syscall-abi.c index d8eeeafb50dc..1e13b7523918 100644 --- a/tools/testing/selftests/arm64/abi/syscall-abi.c +++ b/tools/testing/selftests/arm64/abi/syscall-abi.c @@ -18,7 +18,6 @@

#include "../../kselftest.h"

-#define ARRAY_SIZE(a) (sizeof(a) / sizeof(a[0])) #define NUM_VL ((SVE_VQ_MAX - SVE_VQ_MIN) + 1)

extern void do_syscall(int sve_vl); diff --git a/tools/testing/selftests/arm64/fp/sve-ptrace.c b/tools/testing/selftests/arm64/fp/sve-ptrace.c index a3c1e67441f9..4bd333768cc4 100644 --- a/tools/testing/selftests/arm64/fp/sve-ptrace.c +++ b/tools/testing/selftests/arm64/fp/sve-ptrace.c @@ -21,8 +21,6 @@

#include "../../kselftest.h"

-#define ARRAY_SIZE(a) (sizeof(a) / sizeof(a[0])) - /* <linux/elf.h> and <sys/auxv.h> don't like each other, so: */ #ifndef NT_ARM_SVE #define NT_ARM_SVE 0x405

Provide ABI documentation for SME similar to that for SVE. Due to the very large overlap around streaming SVE mode in both implementation and interfaces documentation for streaming mode SVE is added to the SVE document rather than the SME one.

Signed-off-by: Mark Brown broonie@kernel.org --- Documentation/arm64/index.rst | 1 + Documentation/arm64/sme.rst | 427 ++++++++++++++++++++++++++++++++++ Documentation/arm64/sve.rst | 70 +++++- 3 files changed, 488 insertions(+), 10 deletions(-) create mode 100644 Documentation/arm64/sme.rst

diff --git a/Documentation/arm64/index.rst b/Documentation/arm64/index.rst index 4f840bac083e..ae21f8118830 100644 --- a/Documentation/arm64/index.rst +++ b/Documentation/arm64/index.rst @@ -21,6 +21,7 @@ ARM64 Architecture perf pointer-authentication silicon-errata + sme sve tagged-address-abi tagged-pointers diff --git a/Documentation/arm64/sme.rst b/Documentation/arm64/sme.rst new file mode 100644 index 000000000000..0493838bdc44 --- /dev/null +++ b/Documentation/arm64/sme.rst @@ -0,0 +1,427 @@ +=================================================== +Scalable Matrix Extension support for AArch64 Linux +=================================================== + +This document outlines briefly the interface provided to userspace by Linux in +order to support use of the ARM Scalable Matrix Extension (SME). + +This is an outline of the most important features and issues only and not +intended to be exhaustive. It should be read in conjunction with the SVE +documentation in sve.rst which provides details on the Streaming SVE mode +included in SME. + +This document does not aim to describe the SME architecture or programmer's +model. To aid understanding, a minimal description of relevant programmer's +model features for SME is included in Appendix A. + + +1. General +----------- + +* PSTATE.SM, PSTATE.ZA, the streaming mode vector length, the ZA + register state and TPIDR2_EL0 are tracked per thread. + +* The presence of SME is reported to userspace via HWCAP2_SME in the aux vector + AT_HWCAP2 entry. Presence of this flag implies the presence of the SME + instructions and registers, and the Linux-specific system interfaces + described in this document. SME is reported in /proc/cpuinfo as "sme". + +* Support for the execution of SME instructions in userspace can also be + detected by reading the CPU ID register ID_AA64PFR1_EL1 using an MRS + instruction, and checking that the value of the SME field is nonzero. [3] + + It does not guarantee the presence of the system interfaces described in the + following sections: software that needs to verify that those interfaces are + present must check for HWCAP2_SME instead. + +* There are a number of optional SME features, presence of these is reported + through AT_HWCAP2 through: + + HWCAP2_SME_I16I64 + HWCAP2_SME_F64F64 + HWCAP2_SME_I8I32 + HWCAP2_SME_F16F32 + HWCAP2_SME_B16F32 + HWCAP2_SME_F32F32 + HWCAP2_SME_FA64 + + This list may be extended over time as the SME architecture evolves. + + These extensions are also reported via the CPU ID register ID_AA64SMFR0_EL1, + which userspace can read using an MRS instruction. See elf_hwcaps.txt and + cpu-feature-registers.txt for details. + +* Debuggers should restrict themselves to interacting with the target via the + NT_ARM_SVE, NT_ARM_SSVE and NT_ARM_ZA regsets. The recommended way + of detecting support for these regsets is to connect to a target process + first and then attempt a + + ptrace(PTRACE_GETREGSET, pid, NT_ARM_<regset>, &iov). + +* Whenever ZA register values are exchanged in memory between userspace and + the kernel, the register value is encoded in memory as a series of horizontal + vectors from 0 to VL/8-1 stored in the same endianness invariant format as is + used for SVE vectors. + +* On thread creation TPIDR2_EL0 is preserved unless CLONE_SETTLS is specified, + in which case it is set to 0. + +2. Vector lengths +------------------ + +SME defines a second vector length similar to the SVE vector length which is +controls the size of the streaming mode SVE vectors and the ZA matrix array. +The ZA matrix is square with each side having as many bytes as a SVE vector. + + +3. Sharing of streaming and non-streaming mode SVE state +--------------------------------------------------------- + +It is implementation defined which if any parts of the SVE state are shared +between streaming and non-streaming modes. When switching between modes +via software interfaces such as ptrace if no register content is provided as +part of switching no state will be assumed to be shared and everything will +be zeroed. + + +4. System call behaviour +------------------------- + +* On syscall PSTATE.ZA is preserved, if PSTATE.ZA==1 then the contents of the + ZA matrix are preserved. + +* On syscall PSTATE.SM will be cleared and the SVE registers will be handled + as normal. + +* Neither the SVE registers nor ZA are used to pass arguments to or receive + results from any syscall. + +* On creation fork() or clone() the newly created process will have PSTATE.SM + and PSTATE.ZA cleared. + +* All other SME state of a thread, including the currently configured vector + length, the state of the PR_SME_VL_INHERIT flag, and the deferred vector + length (if any), is preserved across all syscalls, subject to the specific + exceptions for execve() described in section 6. + + +5. Signal handling +------------------- + +* Signal handlers are invoked with streaming mode and ZA disabled. + +* A new signal frame record za_context encodes the ZA register contents on + signal delivery. [1] + +* The signal frame record for ZA always contains basic metadata, in particular + the thread's vector length (in za_context.vl). + +* The ZA matrix may or may not be included in the record, depending on + the value of PSTATE.ZA. The registers are present if and only if: + za_context.head.size >= ZA_SIG_CONTEXT_SIZE(sve_vq_from_vl(za_context.vl)) + in which case PSTATE.ZA == 1. + +* If matrix data is present, the remainder of the record has a vl-dependent + size and layout. Macros ZA_SIG_* are defined [1] to facilitate access to + them. + +* The matrix is stored as a series of horizontal vectors in the same format as + is used for SVE vectors. + +* If the ZA context is too big to fit in sigcontext.__reserved[], then extra + space is allocated on the stack, an extra_context record is written in + __reserved[] referencing this space. za_context is then written in the + extra space. Refer to [1] for further details about this mechanism. + + +5. Signal return +----------------- + +When returning from a signal handler: + +* If there is no za_context record in the signal frame, or if the record is + present but contains no register data as described in the previous section, + then ZA is disabled. + +* If za_context is present in the signal frame and contains matrix data then + PSTATE.ZA is set to 1 and ZA is populated with the specified data. + +* The vector length cannot be changed via signal return. If za_context.vl in + the signal frame does not match the current vector length, the signal return + attempt is treated as illegal, resulting in a forced SIGSEGV. + + +6. prctl extensions +-------------------- + +Some new prctl() calls are added to allow programs to manage the SME vector +length: + +prctl(PR_SME_SET_VL, unsigned long arg) + + Sets the vector length of the calling thread and related flags, where + arg == vl | flags. Other threads of the calling process are unaffected. + + vl is the desired vector length, where sve_vl_valid(vl) must be true. + + flags: + + PR_SME_VL_INHERIT + + Inherit the current vector length across execve(). Otherwise, the + vector length is reset to the system default at execve(). (See + Section 9.) + + PR_SME_SET_VL_ONEXEC + + Defer the requested vector length change until the next execve() + performed by this thread. + + The effect is equivalent to implicit execution of the following + call immediately after the next execve() (if any) by the thread: + + prctl(PR_SME_SET_VL, arg & ~PR_SME_SET_VL_ONEXEC) + + This allows launching of a new program with a different vector + length, while avoiding runtime side effects in the caller. + + Without PR_SME_SET_VL_ONEXEC, the requested change takes effect + immediately. + + + Return value: a nonnegative on success, or a negative value on error: + EINVAL: SME not supported, invalid vector length requested, or + invalid flags. + + + On success: + + * Either the calling thread's vector length or the deferred vector length + to be applied at the next execve() by the thread (dependent on whether + PR_SME_SET_VL_ONEXEC is present in arg), is set to the largest value + supported by the system that is less than or equal to vl. If vl == + SVE_VL_MAX, the value set will be the largest value supported by the + system. + + * Any previously outstanding deferred vector length change in the calling + thread is cancelled. + + * The returned value describes the resulting configuration, encoded as for + PR_SME_GET_VL. The vector length reported in this value is the new + current vector length for this thread if PR_SME_SET_VL_ONEXEC was not + present in arg; otherwise, the reported vector length is the deferred + vector length that will be applied at the next execve() by the calling + thread. + + * Changing the vector length causes all of ZA, P0..P15, FFR and all bits of + Z0..Z31 except for Z0 bits [127:0] .. Z31 bits [127:0] to become + unspecified, including both streaming and non-streaming SVE state. + Calling PR_SME_SET_VL with vl equal to the thread's current vector + length, or calling PR_SME_SET_VL with the PR_SVE_SET_VL_ONEXEC flag, + does not constitute a change to the vector length for this purpose. + + * Changing the vector length causes PSTATE.ZA and PSTATE.SM to be cleared. + Calling PR_SME_SET_VL with vl equal to the thread's current vector + length, or calling PR_SME_SET_VL with the PR_SVE_SET_VL_ONEXEC flag, + does not constitute a change to the vector length for this purpose. + + +prctl(PR_SME_GET_VL) + + Gets the vector length of the calling thread. + + The following flag may be OR-ed into the result: + + PR_SME_VL_INHERIT + + Vector length will be inherited across execve(). + + There is no way to determine whether there is an outstanding deferred + vector length change (which would only normally be the case between a + fork() or vfork() and the corresponding execve() in typical use). + + To extract the vector length from the result, bitwise and it with + PR_SME_VL_LEN_MASK. + + Return value: a nonnegative value on success, or a negative value on error: + EINVAL: SME not supported. + + +7. ptrace extensions +--------------------- + +* A new regset NT_ARM_SSVE is defined for access to streaming mode SVE + state via PTRACE_GETREGSET and PTRACE_SETREGSET, this is documented in + sve.rst. + +* A new regset NT_ARM_ZA is defined for ZA state for access to ZA state via + PTRACE_GETREGSET and PTRACE_SETREGSET. + + Refer to [2] for definitions. + +The regset data starts with struct user_za_header, containing: + + size + + Size of the complete regset, in bytes. + This depends on vl and possibly on other things in the future. + + If a call to PTRACE_GETREGSET requests less data than the value of + size, the caller can allocate a larger buffer and retry in order to + read the complete regset. + + max_size + + Maximum size in bytes that the regset can grow to for the target + thread. The regset won't grow bigger than this even if the target + thread changes its vector length etc. + + vl + + Target thread's current streaming vector length, in bytes. + + max_vl + + Maximum possible streaming vector length for the target thread. + + flags + + Zero or more of the following flags, which have the same + meaning and behaviour as the corresponding PR_SET_VL_* flags: + + SME_PT_VL_INHERIT + + SME_PT_VL_ONEXEC (SETREGSET only). + +* The effects of changing the vector length and/or flags are equivalent to + those documented for PR_SME_SET_VL. + + The caller must make a further GETREGSET call if it needs to know what VL is + actually set by SETREGSET, unless is it known in advance that the requested + VL is supported. + +* The size and layout of the payload depends on the header fields. The + SME_PT_ZA_*() macros are provided to facilitate access to the data. + +* In either case, for SETREGSET it is permissible to omit the payload, in which + case the vector length and flags are changed and PSTATE.ZA is set to 0 + (along with any consequences of those changes). If a payload is provided + then PSTATE.ZA will be set to 1. + +* For SETREGSET, if the requested VL is not supported, the effect will be the + same as if the payload were omitted, except that an EIO error is reported. + No attempt is made to translate the payload data to the correct layout + for the vector length actually set. It is up to the caller to translate the + payload layout for the actual VL and retry. + +* The effect of writing a partial, incomplete payload is unspecified. + + +8. ELF coredump extensions +--------------------------- + +* NT_ARM_SSVE notes will be added to each coredump for + each thread of the dumped process. The contents will be equivalent to the + data that would have been read if a PTRACE_GETREGSET of the corresponding + type were executed for each thread when the coredump was generated. + +* A NT_ARM_ZA note will be added to each coredump for each thread of the + dumped process. The contents will be equivalent to the data that would have + been read if a PTRACE_GETREGSET of NT_ARM_ZA were executed for each thread + when the coredump was generated. + + +9. System runtime configuration +-------------------------------- + +* To mitigate the ABI impact of expansion of the signal frame, a policy + mechanism is provided for administrators, distro maintainers and developers + to set the default vector length for userspace processes: + +/proc/sys/abi/sme_default_vector_length + + Writing the text representation of an integer to this file sets the system + default vector length to the specified value, unless the value is greater + than the maximum vector length supported by the system in which case the + default vector length is set to that maximum. + + The result can be determined by reopening the file and reading its + contents. + + At boot, the default vector length is initially set to 32 or the maximum + supported vector length, whichever is smaller and supported. This + determines the initial vector length of the init process (PID 1). + + Reading this file returns the current system default vector length. + +* At every execve() call, the new vector length of the new process is set to + the system default vector length, unless + + * PR_SME_VL_INHERIT (or equivalently SME_PT_VL_INHERIT) is set for the + calling thread, or + + * a deferred vector length change is pending, established via the + PR_SME_SET_VL_ONEXEC flag (or SME_PT_VL_ONEXEC). + +* Modifying the system default vector length does not affect the vector length + of any existing process or thread that does not make an execve() call. + + +Appendix A. SME programmer's model (informative) +================================================= + +This section provides a minimal description of the additions made by SVE to the +ARMv8-A programmer's model that are relevant to this document. + +Note: This section is for information only and not intended to be complete or +to replace any architectural specification. + +A.1. Registers +--------------- + +In A64 state, SME adds the following: + +* A new mode, streaming mode, in which a subset of the normal FPSIMD and SVE + features are available. When supported EL0 software may enter and leave + streaming mode at any time. + + For best system performance it is strongly encouraged for software to enable + streaming mode only when it is actively being used. + +* A new vector length controlling the size of ZA and the Z registers when in + streaming mode, separately to the vector length used for SVE when not in + streaming mode. There is no requirement that either the currently selected + vector length or the set of vector lengths supported for the two modes in + a given system have any relationship. The streaming mode vector length + is referred to as SVL. + +* A new ZA matrix register. This is a square matrix of SVLxSVL bits. Most + operations on ZA require that streaming mode be enabled but ZA can be + enabled without streaming mode in order to load, save and retain data. + + For best system performance it is strongly encouraged for software to enable + ZA only when it is actively being used. + +* Two new 1 bit fields in PSTATE which may be controlled via the SMSTART and + SMSTOP instructions or by access to the SVCR system register: + + * PSTATE.ZA, if this is 1 then the ZA matrix is accessible and has valid + data while if it is 0 then ZA can not be accessed. When PSTATE.ZA is + changed from 0 to 1 all bits in ZA are cleared. + + * PSTATE.SM, if this is 1 then the PE is in streaming mode. When the value + of PSTATE.SM is changed then it is implementation defined if the subset + of the floating point register bits valid in both modes may be retained. + Any other bits will be cleared. + + +References +========== + +[1] arch/arm64/include/uapi/asm/sigcontext.h + AArch64 Linux signal ABI definitions + +[2] arch/arm64/include/uapi/asm/ptrace.h + AArch64 Linux ptrace ABI definitions + +[3] Documentation/arm64/cpu-feature-registers.rst diff --git a/Documentation/arm64/sve.rst b/Documentation/arm64/sve.rst index 9d9a4de5bc34..93c2c2990584 100644 --- a/Documentation/arm64/sve.rst +++ b/Documentation/arm64/sve.rst @@ -7,7 +7,9 @@ Author: Dave Martin Dave.Martin@arm.com Date: 4 August 2017

This document outlines briefly the interface provided to userspace by Linux in -order to support use of the ARM Scalable Vector Extension (SVE). +order to support use of the ARM Scalable Vector Extension (SVE), including +interactions with Streaming SVE mode added by the Scalable Matrix Extension +(SME).

This is an outline of the most important features and issues only and not intended to be exhaustive. @@ -23,6 +25,10 @@ model features for SVE is included in Appendix A. * SVE registers Z0..Z31, P0..P15 and FFR and the current vector length VL, are tracked per-thread.

+* In streaming mode FFR is not accessible unless HWCAP2_SME_FA64 is present + in the system, when it is not supported and these interfaces are used to + access streaming mode FFR is read and written as zero. + * The presence of SVE is reported to userspace via HWCAP_SVE in the aux vector AT_HWCAP entry. Presence of this flag implies the presence of the SVE instructions and registers, and the Linux-specific system interfaces @@ -53,10 +59,19 @@ model features for SVE is included in Appendix A. which userspace can read using an MRS instruction. See elf_hwcaps.txt and cpu-feature-registers.txt for details.

+* On hardware that supports the SME extensions, HWCAP2_SME will also be + reported in the AT_HWCAP2 aux vector entry. Among other things SME adds + streaming mode which provides a subset of the SVE feature set using a + separate SME vector length and the same Z/V registers. See sme.rst + for more details. + * Debuggers should restrict themselves to interacting with the target via the NT_ARM_SVE regset. The recommended way of detecting support for this regset is to connect to a target process first and then attempt a - ptrace(PTRACE_GETREGSET, pid, NT_ARM_SVE, &iov). + ptrace(PTRACE_GETREGSET, pid, NT_ARM_SVE, &iov). Note that when SME is + present and streaming SVE mode is in use the FPSIMD subset of registers + will be read via NT_ARM_SVE and NT_ARM_SVE writes will exit streaming mode + in the target.

* Whenever SVE scalable register values (Zn, Pn, FFR) are exchanged in memory between userspace and the kernel, the register value is encoded in memory in @@ -126,6 +141,11 @@ the SVE instruction set architecture. are only present in fpsimd_context. For convenience, the content of V0..V31 is duplicated between sve_context and fpsimd_context.

+* The record contains a flag field which includes a flag SVE_SIG_FLAG_SM which + if set indicates that the thread is in streaming mode and the vector length + and register data (if present) describe the streaming SVE data and vector + length. + * The signal frame record for SVE always contains basic metadata, in particular the thread's vector length (in sve_context.vl).

@@ -170,6 +190,11 @@ When returning from a signal handler: the signal frame does not match the current vector length, the signal return attempt is treated as illegal, resulting in a forced SIGSEGV.

+* It is permitted to enter or leave streaming mode by setting or clearing + the SVE_SIG_FLAG_SM flag but applications should take care to ensure that + when doing so sve_context.vl and any register data are appropriate for the + vector length in the new mode. +

6. prctl extensions -------------------- @@ -265,8 +290,14 @@ prctl(PR_SVE_GET_VL) 7. ptrace extensions ---------------------

-* A new regset NT_ARM_SVE is defined for use with PTRACE_GETREGSET and - PTRACE_SETREGSET. +* New regsets NT_ARM_SVE and NT_ARM_SSVE are defined for use with + PTRACE_GETREGSET and PTRACE_SETREGSET. NT_ARM_SSVE describes the + streaming mode SVE registers and NT_ARM_SVE describes the + non-streaming mode SVE registers. + + In this description a register set is referred to as being "live" when + the target is in the appropriate streaming or non-streaming mode and is + using data beyond the subset shared with the FPSIMD Vn registers.

Refer to [2] for definitions.

@@ -297,7 +328,7 @@ The regset data starts with struct user_sve_header, containing:

flags

- either + at most one of

SVE_PT_REGS_FPSIMD

@@ -331,6 +362,10 @@ The regset data starts with struct user_sve_header, containing:

SVE_PT_VL_ONEXEC (SETREGSET only).

+ If neither FPSIMD nor SVE flags are provided then no register + payload is available, this is only possible when SME is implemented. + + * The effects of changing the vector length and/or flags are equivalent to those documented for PR_SVE_SET_VL.

@@ -346,6 +381,13 @@ The regset data starts with struct user_sve_header, containing: case only the vector length and flags are changed (along with any consequences of those changes).

+* In systems supporting SME when in streaming mode a GETREGSET for + NT_REG_SVE will return only the user_sve_header with no register data, + similarly a GETREGSET for NT_REG_SSVE will not return any register data + when not in streaming mode. + +* A GETREGSET for NT_ARM_SSVE will never return SVE_PT_REGS_FPSIMD. + * For SETREGSET, if an SVE_PT_REGS_SVE payload is present and the requested VL is not supported, the effect will be the same as if the payload were omitted, except that an EIO error is reported. No @@ -355,17 +397,25 @@ The regset data starts with struct user_sve_header, containing: unspecified. It is up to the caller to translate the payload layout for the actual VL and retry.

+* Where SME is implemented it is not possible to GETREGSET the register + state for normal SVE when in streaming mode, nor the streaming mode + register state when in normal mode, regardless of the implementation defined + behaviour of the hardware for sharing data between the two modes. + +* Any SETREGSET of NT_ARM_SVE will exit streaming mode if the target was in + streaming mode and any SETREGSET of NT_ARM_SSVE will enter streaming mode + if the target was not in streaming mode. + * The effect of writing a partial, incomplete payload is unspecified.

8. ELF coredump extensions ---------------------------

-* A NT_ARM_SVE note will be added to each coredump for each thread of the - dumped process. The contents will be equivalent to the data that would have - been read if a PTRACE_GETREGSET of NT_ARM_SVE were executed for each thread - when the coredump was generated. - +* NT_ARM_SVE and NT_ARM_SSVE notes will be added to each coredump for + each thread of the dumped process. The contents will be equivalent to the + data that would have been read if a PTRACE_GETREGSET of the corresponding + type were executed for each thread when the coredump was generated.

9. System runtime configuration --------------------------------

The arm64 Scalable Matrix Extension (SME) adds some new system registers, fields in existing system registers and exception syndromes. This patch adds definitions for these for use in future patches implementing support for this extension.

Since SME will be the first user of FEAT_HCX in the kernel also include the definitions for enumerating it and the HCRX system register it adds.

Signed-off-by: Mark Brown broonie@kernel.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/esr.h | 12 +++++- arch/arm64/include/asm/kvm_arm.h | 1 + arch/arm64/include/asm/sysreg.h | 67 ++++++++++++++++++++++++++++++++ arch/arm64/kernel/traps.c | 1 + 4 files changed, 80 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/esr.h b/arch/arm64/include/asm/esr.h index d52a0b269ee8..43872e0cfd1e 100644 --- a/arch/arm64/include/asm/esr.h +++ b/arch/arm64/include/asm/esr.h @@ -37,7 +37,8 @@ #define ESR_ELx_EC_ERET (0x1a) /* EL2 only */ /* Unallocated EC: 0x1B */ #define ESR_ELx_EC_FPAC (0x1C) /* EL1 and above */ -/* Unallocated EC: 0x1D - 0x1E */ +#define ESR_ELx_EC_SME (0x1D) +/* Unallocated EC: 0x1E */ #define ESR_ELx_EC_IMP_DEF (0x1f) /* EL3 only */ #define ESR_ELx_EC_IABT_LOW (0x20) #define ESR_ELx_EC_IABT_CUR (0x21) @@ -327,6 +328,15 @@ #define ESR_ELx_CP15_32_ISS_SYS_CNTFRQ (ESR_ELx_CP15_32_ISS_SYS_VAL(0, 0, 14, 0) |\ ESR_ELx_CP15_32_ISS_DIR_READ)

+/* + * ISS values for SME traps + */ + +#define ESR_ELx_SME_ISS_SME_DISABLED 0 +#define ESR_ELx_SME_ISS_ILL 1 +#define ESR_ELx_SME_ISS_SM_DISABLED 2 +#define ESR_ELx_SME_ISS_ZA_DISABLED 3 + #ifndef __ASSEMBLY__ #include <asm/types.h>

diff --git a/arch/arm64/include/asm/kvm_arm.h b/arch/arm64/include/asm/kvm_arm.h index 1767ded83888..13ae232ec4a1 100644 --- a/arch/arm64/include/asm/kvm_arm.h +++ b/arch/arm64/include/asm/kvm_arm.h @@ -279,6 +279,7 @@ #define CPTR_EL2_TCPAC (1U << 31) #define CPTR_EL2_TAM (1 << 30) #define CPTR_EL2_TTA (1 << 20) +#define CPTR_EL2_TSM (1 << 12) #define CPTR_EL2_TFP (1 << CPTR_EL2_TFP_SHIFT) #define CPTR_EL2_TZ (1 << 8) #define CPTR_NVHE_EL2_RES1 0x000032ff /* known RES1 bits in CPTR_EL2 (nVHE) */ diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h index e66dd9ebc337..77d1d4cd8e2f 100644 --- a/arch/arm64/include/asm/sysreg.h +++ b/arch/arm64/include/asm/sysreg.h @@ -118,6 +118,10 @@ * System registers, organised loosely by encoding but grouped together * where the architected name contains an index. e.g. ID_MMFR<n>_EL1. */ +#define SYS_SVCR_SMSTOP_SM_EL0 sys_reg(0, 3, 4, 2, 3) +#define SYS_SVCR_SMSTART_SM_EL0 sys_reg(0, 3, 4, 3, 3) +#define SYS_SVCR_SMSTOP_SMZA_EL0 sys_reg(0, 3, 4, 6, 3) + #define SYS_OSDTRRX_EL1 sys_reg(2, 0, 0, 0, 2) #define SYS_MDCCINT_EL1 sys_reg(2, 0, 0, 2, 0) #define SYS_MDSCR_EL1 sys_reg(2, 0, 0, 2, 2) @@ -173,6 +177,7 @@ #define SYS_ID_AA64PFR0_EL1 sys_reg(3, 0, 0, 4, 0) #define SYS_ID_AA64PFR1_EL1 sys_reg(3, 0, 0, 4, 1) #define SYS_ID_AA64ZFR0_EL1 sys_reg(3, 0, 0, 4, 4) +#define SYS_ID_AA64SMFR0_EL1 sys_reg(3, 0, 0, 4, 5)

#define SYS_ID_AA64DFR0_EL1 sys_reg(3, 0, 0, 5, 0) #define SYS_ID_AA64DFR1_EL1 sys_reg(3, 0, 0, 5, 1) @@ -196,6 +201,8 @@

#define SYS_ZCR_EL1 sys_reg(3, 0, 1, 2, 0) #define SYS_TRFCR_EL1 sys_reg(3, 0, 1, 2, 1) +#define SYS_SMPRI_EL1 sys_reg(3, 0, 1, 2, 4) +#define SYS_SMCR_EL1 sys_reg(3, 0, 1, 2, 6)

#define SYS_TTBR0_EL1 sys_reg(3, 0, 2, 0, 0) #define SYS_TTBR1_EL1 sys_reg(3, 0, 2, 0, 1) @@ -388,6 +395,8 @@ #define TRBIDR_ALIGN_MASK GENMASK(3, 0) #define TRBIDR_ALIGN_SHIFT 0

+#define SMPRI_EL1_PRIORITY_MASK 0xf + #define SYS_PMINTENSET_EL1 sys_reg(3, 0, 9, 14, 1) #define SYS_PMINTENCLR_EL1 sys_reg(3, 0, 9, 14, 2)

@@ -443,8 +452,13 @@ #define SYS_CCSIDR_EL1 sys_reg(3, 1, 0, 0, 0) #define SYS_CLIDR_EL1 sys_reg(3, 1, 0, 0, 1) #define SYS_GMID_EL1 sys_reg(3, 1, 0, 0, 4) +#define SYS_SMIDR_EL1 sys_reg(3, 1, 0, 0, 6) #define SYS_AIDR_EL1 sys_reg(3, 1, 0, 0, 7)

+#define SYS_SMIDR_EL1_IMPLEMENTER_SHIFT 24 +#define SYS_SMIDR_EL1_SMPS_SHIFT 15 +#define SYS_SMIDR_EL1_AFFINITY_SHIFT 0 + #define SYS_CSSELR_EL1 sys_reg(3, 2, 0, 0, 0)

#define SYS_CTR_EL0 sys_reg(3, 3, 0, 0, 1) @@ -453,6 +467,10 @@ #define SYS_RNDR_EL0 sys_reg(3, 3, 2, 4, 0) #define SYS_RNDRRS_EL0 sys_reg(3, 3, 2, 4, 1)

+#define SYS_SVCR_EL0 sys_reg(3, 3, 4, 2, 2) +#define SYS_SVCR_EL0_ZA_MASK 2 +#define SYS_SVCR_EL0_SM_MASK 1 + #define SYS_PMCR_EL0 sys_reg(3, 3, 9, 12, 0) #define SYS_PMCNTENSET_EL0 sys_reg(3, 3, 9, 12, 1) #define SYS_PMCNTENCLR_EL0 sys_reg(3, 3, 9, 12, 2) @@ -469,6 +487,7 @@

#define SYS_TPIDR_EL0 sys_reg(3, 3, 13, 0, 2) #define SYS_TPIDRRO_EL0 sys_reg(3, 3, 13, 0, 3) +#define SYS_TPIDR2_EL0 sys_reg(3, 3, 13, 0, 5)

#define SYS_SCXTNUM_EL0 sys_reg(3, 3, 13, 0, 7)

@@ -538,6 +557,9 @@ #define SYS_HFGITR_EL2 sys_reg(3, 4, 1, 1, 6) #define SYS_ZCR_EL2 sys_reg(3, 4, 1, 2, 0) #define SYS_TRFCR_EL2 sys_reg(3, 4, 1, 2, 1) +#define SYS_HCRX_EL2 sys_reg(3, 4, 1, 2, 2) +#define SYS_SMPRIMAP_EL2 sys_reg(3, 4, 1, 2, 5) +#define SYS_SMCR_EL2 sys_reg(3, 4, 1, 2, 6) #define SYS_DACR32_EL2 sys_reg(3, 4, 3, 0, 0) #define SYS_HDFGRTR_EL2 sys_reg(3, 4, 3, 1, 4) #define SYS_HDFGWTR_EL2 sys_reg(3, 4, 3, 1, 5) @@ -597,6 +619,7 @@ #define SYS_SCTLR_EL12 sys_reg(3, 5, 1, 0, 0) #define SYS_CPACR_EL12 sys_reg(3, 5, 1, 0, 2) #define SYS_ZCR_EL12 sys_reg(3, 5, 1, 2, 0) +#define SYS_SMCR_EL12 sys_reg(3, 5, 1, 2, 6) #define SYS_TTBR0_EL12 sys_reg(3, 5, 2, 0, 0) #define SYS_TTBR1_EL12 sys_reg(3, 5, 2, 0, 1) #define SYS_TCR_EL12 sys_reg(3, 5, 2, 0, 2) @@ -620,6 +643,7 @@ #define SYS_CNTV_CVAL_EL02 sys_reg(3, 5, 14, 3, 2)

/* Common SCTLR_ELx flags. */ +#define SCTLR_ELx_ENTP2 (BIT(60)) #define SCTLR_ELx_DSSBS (BIT(44)) #define SCTLR_ELx_ATA (BIT(43))

@@ -815,6 +839,7 @@ #define ID_AA64PFR0_ELx_32BIT_64BIT 0x2

/* id_aa64pfr1 */ +#define ID_AA64PFR1_SME_SHIFT 24 #define ID_AA64PFR1_MPAMFRAC_SHIFT 16 #define ID_AA64PFR1_RASFRAC_SHIFT 12 #define ID_AA64PFR1_MTE_SHIFT 8 @@ -825,6 +850,7 @@ #define ID_AA64PFR1_SSBS_PSTATE_ONLY 1 #define ID_AA64PFR1_SSBS_PSTATE_INSNS 2 #define ID_AA64PFR1_BT_BTI 0x1 +#define ID_AA64PFR1_SME 1

#define ID_AA64PFR1_MTE_NI 0x0 #define ID_AA64PFR1_MTE_EL0 0x1 @@ -853,6 +879,23 @@ #define ID_AA64ZFR0_AES_PMULL 0x2 #define ID_AA64ZFR0_SVEVER_SVE2 0x1

+/* id_aa64smfr0 */ +#define ID_AA64SMFR0_FA64_SHIFT 63 +#define ID_AA64SMFR0_I16I64_SHIFT 52 +#define ID_AA64SMFR0_F64F64_SHIFT 48 +#define ID_AA64SMFR0_I8I32_SHIFT 36 +#define ID_AA64SMFR0_F16F32_SHIFT 35 +#define ID_AA64SMFR0_B16F32_SHIFT 34 +#define ID_AA64SMFR0_F32F32_SHIFT 32 + +#define ID_AA64SMFR0_FA64 0x1 +#define ID_AA64SMFR0_I16I64 0x4 +#define ID_AA64SMFR0_F64F64 0x1 +#define ID_AA64SMFR0_I8I32 0x4 +#define ID_AA64SMFR0_F16F32 0x1 +#define ID_AA64SMFR0_B16F32 0x1 +#define ID_AA64SMFR0_F32F32 0x1 + /* id_aa64mmfr0 */ #define ID_AA64MMFR0_ECV_SHIFT 60 #define ID_AA64MMFR0_FGT_SHIFT 56 @@ -904,6 +947,7 @@ #endif

/* id_aa64mmfr1 */ +#define ID_AA64MMFR1_HCX_SHIFT 40 #define ID_AA64MMFR1_AFP_SHIFT 44 #define ID_AA64MMFR1_ETS_SHIFT 36 #define ID_AA64MMFR1_TWED_SHIFT 32 @@ -1097,9 +1141,24 @@ #define ZCR_ELx_LEN_SIZE 9 #define ZCR_ELx_LEN_MASK 0x1ff

+#define SMCR_ELx_FA64_SHIFT 31 +#define SMCR_ELx_FA64_MASK (1 << SMCR_ELx_FA64_SHIFT) + +/* + * The SMCR_ELx_LEN_* definitions intentionally include bits [8:4] which + * are reserved by the SME architecture for future expansion of the LEN + * field, with compatible semantics. + */ +#define SMCR_ELx_LEN_SHIFT 0 +#define SMCR_ELx_LEN_SIZE 9 +#define SMCR_ELx_LEN_MASK 0x1ff + #define CPACR_EL1_FPEN_EL1EN (BIT(20)) /* enable EL1 access */ #define CPACR_EL1_FPEN_EL0EN (BIT(21)) /* enable EL0 access, if EL1EN set */

+#define CPACR_EL1_SMEN_EL1EN (BIT(24)) /* enable EL1 access */ +#define CPACR_EL1_SMEN_EL0EN (BIT(25)) /* enable EL0 access, if EL1EN set */ + #define CPACR_EL1_ZEN_EL1EN (BIT(16)) /* enable EL1 access */ #define CPACR_EL1_ZEN_EL0EN (BIT(17)) /* enable EL0 access, if EL1EN set */

@@ -1152,6 +1211,8 @@ #define TRFCR_ELx_ExTRE BIT(1) #define TRFCR_ELx_E0TRE BIT(0)

+/* HCRX_EL2 definitions */ +#define HCRX_EL2_SMPME_MASK (1 << 5)

/* GIC Hypervisor interface registers */ /* ICH_MISR_EL2 bit definitions */ @@ -1215,6 +1276,12 @@ #define ICH_VTR_TDS_SHIFT 19 #define ICH_VTR_TDS_MASK (1 << ICH_VTR_TDS_SHIFT)

+/* HFG[WR]TR_EL2 bit definitions */ +#define HFGxTR_EL2_nTPIDR2_EL0_SHIFT 55 +#define HFGxTR_EL2_nTPIDR2_EL0_MASK BIT_MASK(HFGxTR_EL2_nTPIDR2_EL0_SHIFT) +#define HFGxTR_EL2_nSMPRI_EL1_SHIFT 54 +#define HFGxTR_EL2_nSMPRI_EL1_MASK BIT_MASK(HFGxTR_EL2_nSMPRI_EL1_SHIFT) + #define ARM64_FEATURE_FIELD_BITS 4

/* Create a mask for the feature bits of the specified feature. */ diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c index 70fc42470f13..88effb25c399 100644 --- a/arch/arm64/kernel/traps.c +++ b/arch/arm64/kernel/traps.c @@ -822,6 +822,7 @@ static const char *esr_class_str[] = { [ESR_ELx_EC_SVE] = "SVE", [ESR_ELx_EC_ERET] = "ERET/ERETAA/ERETAB", [ESR_ELx_EC_FPAC] = "FPAC", + [ESR_ELx_EC_SME] = "SME", [ESR_ELx_EC_IMP_DEF] = "EL3 IMP DEF", [ESR_ELx_EC_IABT_LOW] = "IABT (lower EL)", [ESR_ELx_EC_IABT_CUR] = "IABT (current EL)",

This patch introduces basic cpufeature support for discovering the presence of the Scalable Matrix Extension.

Signed-off-by: Mark Brown broonie@kernel.org --- Documentation/arm64/elf_hwcaps.rst | 33 ++++++++++++++++ arch/arm64/include/asm/cpu.h | 1 + arch/arm64/include/asm/cpufeature.h | 12 ++++++ arch/arm64/include/asm/fpsimd.h | 2 + arch/arm64/include/asm/hwcap.h | 8 ++++ arch/arm64/include/uapi/asm/hwcap.h | 8 ++++ arch/arm64/kernel/cpufeature.c | 59 +++++++++++++++++++++++++++++ arch/arm64/kernel/cpuinfo.c | 9 +++++ arch/arm64/kernel/fpsimd.c | 30 +++++++++++++++ arch/arm64/tools/cpucaps | 2 + 10 files changed, 164 insertions(+)

diff --git a/Documentation/arm64/elf_hwcaps.rst b/Documentation/arm64/elf_hwcaps.rst index b72ff17d600a..5626cf208000 100644 --- a/Documentation/arm64/elf_hwcaps.rst +++ b/Documentation/arm64/elf_hwcaps.rst @@ -259,6 +259,39 @@ HWCAP2_RPRES

Functionality implied by ID_AA64ISAR2_EL1.RPRES == 0b0001.

+HWCAP2_SME + + Functionality implied by ID_AA64PFR1_EL1.SME == 0b0001, as described + by Documentation/arm64/sme.rst. + +HWCAP2_SME_I16I64 + + Functionality implied by ID_AA64SMFR0_EL1.I16I64 == 0b1111. + +HWCAP2_SME_F64F64 + + Functionality implied by ID_AA64SMFR0_EL1.F64F64 == 0b1. + +HWCAP2_SME_I8I32 + + Functionality implied by ID_AA64SMFR0_EL1.I8I32 == 0b1111. + +HWCAP2_SME_F16F32 + + Functionality implied by ID_AA64SMFR0_EL1.F16F32 == 0b1. + +HWCAP2_SME_B16F32 + + Functionality implied by ID_AA64SMFR0_EL1.B16F32 == 0b1. + +HWCAP2_SME_F32F32 + + Functionality implied by ID_AA64SMFR0_EL1.F32F32 == 0b1. + +HWCAP2_SME_FA64 + + Functionality implied by ID_AA64SMFR0_EL1.FA64 == 0b1. + 4. Unused AT_HWCAP bits -----------------------

diff --git a/arch/arm64/include/asm/cpu.h b/arch/arm64/include/asm/cpu.h index a58e366f0b07..d08062bcb9c1 100644 --- a/arch/arm64/include/asm/cpu.h +++ b/arch/arm64/include/asm/cpu.h @@ -58,6 +58,7 @@ struct cpuinfo_arm64 { u64 reg_id_aa64pfr0; u64 reg_id_aa64pfr1; u64 reg_id_aa64zfr0; + u64 reg_id_aa64smfr0;

struct cpuinfo_32bit aarch32;

diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h index 2728abd9cae4..f93b1442143f 100644 --- a/arch/arm64/include/asm/cpufeature.h +++ b/arch/arm64/include/asm/cpufeature.h @@ -728,6 +728,18 @@ static __always_inline bool system_supports_sve(void) cpus_have_const_cap(ARM64_SVE); }

+static __always_inline bool system_supports_sme(void) +{ + return IS_ENABLED(CONFIG_ARM64_SME) && + cpus_have_const_cap(ARM64_SME); +} + +static __always_inline bool system_supports_fa64(void) +{ + return IS_ENABLED(CONFIG_ARM64_SME) && + cpus_have_const_cap(ARM64_SME_FA64); +} + static __always_inline bool system_supports_cnp(void) { return IS_ENABLED(CONFIG_ARM64_CNP) && diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 6e2dc9dcbf49..2e8ef00e7520 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -74,6 +74,8 @@ extern void sve_set_vq(unsigned long vq_minus_1);

struct arm64_cpu_capabilities; extern void sve_kernel_enable(const struct arm64_cpu_capabilities *__unused); +extern void sme_kernel_enable(const struct arm64_cpu_capabilities *__unused); +extern void fa64_kernel_enable(const struct arm64_cpu_capabilities *__unused);

extern u64 read_zcr_features(void);

diff --git a/arch/arm64/include/asm/hwcap.h b/arch/arm64/include/asm/hwcap.h index f68fbb207473..76d9999527c5 100644 --- a/arch/arm64/include/asm/hwcap.h +++ b/arch/arm64/include/asm/hwcap.h @@ -108,6 +108,14 @@ #define KERNEL_HWCAP_ECV __khwcap2_feature(ECV) #define KERNEL_HWCAP_AFP __khwcap2_feature(AFP) #define KERNEL_HWCAP_RPRES __khwcap2_feature(RPRES) +#define KERNEL_HWCAP_SME __khwcap2_feature(SME) +#define KERNEL_HWCAP_SME_I16I64 __khwcap2_feature(SME_I16I64) +#define KERNEL_HWCAP_SME_F64F64 __khwcap2_feature(SME_F64F64) +#define KERNEL_HWCAP_SME_I8I32 __khwcap2_feature(SME_I8I32) +#define KERNEL_HWCAP_SME_F16F32 __khwcap2_feature(SME_F16F32) +#define KERNEL_HWCAP_SME_B16F32 __khwcap2_feature(SME_B16F32) +#define KERNEL_HWCAP_SME_F32F32 __khwcap2_feature(SME_F32F32) +#define KERNEL_HWCAP_SME_FA64 __khwcap2_feature(SME_FA64)

/* * This yields a mask that user programs can use to figure out what diff --git a/arch/arm64/include/uapi/asm/hwcap.h b/arch/arm64/include/uapi/asm/hwcap.h index f03731847d9d..60de5626f8fb 100644 --- a/arch/arm64/include/uapi/asm/hwcap.h +++ b/arch/arm64/include/uapi/asm/hwcap.h @@ -78,5 +78,13 @@ #define HWCAP2_ECV (1 << 19) #define HWCAP2_AFP (1 << 20) #define HWCAP2_RPRES (1 << 21) +#define HWCAP2_SME (1 << 22) +#define HWCAP2_SME_I16I64 (1 << 23) +#define HWCAP2_SME_F64F64 (1 << 24) +#define HWCAP2_SME_I8I32 (1 << 25) +#define HWCAP2_SME_F16F32 (1 << 26) +#define HWCAP2_SME_B16F32 (1 << 27) +#define HWCAP2_SME_F32F32 (1 << 28) +#define HWCAP2_SME_FA64 (1 << 29)

#endif /* _UAPI__ASM_HWCAP_H */ diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index 64a748c2b351..5bfd440edd75 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -251,6 +251,8 @@ static const struct arm64_ftr_bits ftr_id_aa64pfr0[] = { };

static const struct arm64_ftr_bits ftr_id_aa64pfr1[] = { + ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_SME), + FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR1_SME_SHIFT, 4, 0), ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR1_MPAMFRAC_SHIFT, 4, 0), ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64PFR1_RASFRAC_SHIFT, 4, 0), ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_MTE), @@ -283,6 +285,24 @@ static const struct arm64_ftr_bits ftr_id_aa64zfr0[] = { ARM64_FTR_END, };

+static const struct arm64_ftr_bits ftr_id_aa64smfr0[] = { + ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_SME), + FTR_STRICT, FTR_EXACT, ID_AA64SMFR0_FA64_SHIFT, 1, 0), + ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_SME), + FTR_STRICT, FTR_EXACT, ID_AA64SMFR0_I16I64_SHIFT, 4, 0), + ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_SME), + FTR_STRICT, FTR_EXACT, ID_AA64SMFR0_F64F64_SHIFT, 1, 0), + ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_SME), + FTR_STRICT, FTR_EXACT, ID_AA64SMFR0_I8I32_SHIFT, 4, 0), + ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_SME), + FTR_STRICT, FTR_EXACT, ID_AA64SMFR0_F16F32_SHIFT, 1, 0), + ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_SME), + FTR_STRICT, FTR_EXACT, ID_AA64SMFR0_B16F32_SHIFT, 1, 0), + ARM64_FTR_BITS(FTR_VISIBLE_IF_IS_ENABLED(CONFIG_ARM64_SME), + FTR_STRICT, FTR_EXACT, ID_AA64SMFR0_F32F32_SHIFT, 1, 0), + ARM64_FTR_END, +}; + static const struct arm64_ftr_bits ftr_id_aa64mmfr0[] = { ARM64_FTR_BITS(FTR_VISIBLE, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_ECV_SHIFT, 4, 0), ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_FGT_SHIFT, 4, 0), @@ -634,6 +654,7 @@ static const struct __ftr_reg_entry { ARM64_FTR_REG_OVERRIDE(SYS_ID_AA64PFR1_EL1, ftr_id_aa64pfr1, &id_aa64pfr1_override), ARM64_FTR_REG(SYS_ID_AA64ZFR0_EL1, ftr_id_aa64zfr0), + ARM64_FTR_REG(SYS_ID_AA64SMFR0_EL1, ftr_id_aa64smfr0),

/* Op1 = 0, CRn = 0, CRm = 5 */ ARM64_FTR_REG(SYS_ID_AA64DFR0_EL1, ftr_id_aa64dfr0), @@ -947,6 +968,7 @@ void __init init_cpu_features(struct cpuinfo_arm64 *info) init_cpu_ftr_reg(SYS_ID_AA64PFR0_EL1, info->reg_id_aa64pfr0); init_cpu_ftr_reg(SYS_ID_AA64PFR1_EL1, info->reg_id_aa64pfr1); init_cpu_ftr_reg(SYS_ID_AA64ZFR0_EL1, info->reg_id_aa64zfr0); + init_cpu_ftr_reg(SYS_ID_AA64SMFR0_EL1, info->reg_id_aa64smfr0);

if (id_aa64pfr0_32bit_el0(info->reg_id_aa64pfr0)) init_32bit_cpu_features(&info->aarch32); @@ -2403,6 +2425,33 @@ static const struct arm64_cpu_capabilities arm64_features[] = { .matches = has_cpuid_feature, .min_field_value = 1, }, +#ifdef CONFIG_ARM64_SME + { + .desc = "Scalable Matrix Extension", + .type = ARM64_CPUCAP_SYSTEM_FEATURE, + .capability = ARM64_SME, + .sys_reg = SYS_ID_AA64PFR1_EL1, + .sign = FTR_UNSIGNED, + .field_pos = ID_AA64PFR1_SME_SHIFT, + .field_width = 4, + .min_field_value = ID_AA64PFR1_SME, + .matches = has_cpuid_feature, + .cpu_enable = sme_kernel_enable, + }, + /* FA64 should be sorted after the base SME capability */ + { + .desc = "FA64", + .type = ARM64_CPUCAP_SYSTEM_FEATURE, + .capability = ARM64_SME_FA64, + .sys_reg = SYS_ID_AA64SMFR0_EL1, + .sign = FTR_UNSIGNED, + .field_pos = ID_AA64SMFR0_FA64_SHIFT, + .field_width = 1, + .min_field_value = ID_AA64SMFR0_FA64, + .matches = has_cpuid_feature, + .cpu_enable = fa64_kernel_enable, + }, +#endif /* CONFIG_ARM64_SME */ {}, };

@@ -2527,6 +2576,16 @@ static const struct arm64_cpu_capabilities arm64_elf_hwcaps[] = { HWCAP_CAP(SYS_ID_AA64MMFR0_EL1, ID_AA64MMFR0_ECV_SHIFT, 4, FTR_UNSIGNED, 1, CAP_HWCAP, KERNEL_HWCAP_ECV), HWCAP_CAP(SYS_ID_AA64MMFR1_EL1, ID_AA64MMFR1_AFP_SHIFT, 4, FTR_UNSIGNED, 1, CAP_HWCAP, KERNEL_HWCAP_AFP), HWCAP_CAP(SYS_ID_AA64ISAR2_EL1, ID_AA64ISAR2_RPRES_SHIFT, 4, FTR_UNSIGNED, 1, CAP_HWCAP, KERNEL_HWCAP_RPRES), +#ifdef CONFIG_ARM64_SME + HWCAP_CAP(SYS_ID_AA64PFR1_EL1, ID_AA64PFR1_SME_SHIFT, 4, FTR_UNSIGNED, ID_AA64PFR1_SME, CAP_HWCAP, KERNEL_HWCAP_SME), + HWCAP_CAP(SYS_ID_AA64SMFR0_EL1, ID_AA64SMFR0_FA64_SHIFT, 1, FTR_UNSIGNED, ID_AA64SMFR0_FA64, CAP_HWCAP, KERNEL_HWCAP_SME_FA64), + HWCAP_CAP(SYS_ID_AA64SMFR0_EL1, ID_AA64SMFR0_I16I64_SHIFT, 4, FTR_UNSIGNED, ID_AA64SMFR0_I16I64, CAP_HWCAP, KERNEL_HWCAP_SME_I16I64), + HWCAP_CAP(SYS_ID_AA64SMFR0_EL1, ID_AA64SMFR0_F64F64_SHIFT, 1, FTR_UNSIGNED, ID_AA64SMFR0_F64F64, CAP_HWCAP, KERNEL_HWCAP_SME_F64F64), + HWCAP_CAP(SYS_ID_AA64SMFR0_EL1, ID_AA64SMFR0_I8I32_SHIFT, 4, FTR_UNSIGNED, ID_AA64SMFR0_I8I32, CAP_HWCAP, KERNEL_HWCAP_SME_I8I32), + HWCAP_CAP(SYS_ID_AA64SMFR0_EL1, ID_AA64SMFR0_F16F32_SHIFT, 1, FTR_UNSIGNED, ID_AA64SMFR0_F16F32, CAP_HWCAP, KERNEL_HWCAP_SME_F16F32), + HWCAP_CAP(SYS_ID_AA64SMFR0_EL1, ID_AA64SMFR0_B16F32_SHIFT, 1, FTR_UNSIGNED, ID_AA64SMFR0_B16F32, CAP_HWCAP, KERNEL_HWCAP_SME_B16F32), + HWCAP_CAP(SYS_ID_AA64SMFR0_EL1, ID_AA64SMFR0_F32F32_SHIFT, 1, FTR_UNSIGNED, ID_AA64SMFR0_F32F32, CAP_HWCAP, KERNEL_HWCAP_SME_F32F32), +#endif /* CONFIG_ARM64_SME */ {}, };

diff --git a/arch/arm64/kernel/cpuinfo.c b/arch/arm64/kernel/cpuinfo.c index 591c18a889a5..33ec182e872e 100644 --- a/arch/arm64/kernel/cpuinfo.c +++ b/arch/arm64/kernel/cpuinfo.c @@ -97,6 +97,14 @@ static const char *const hwcap_str[] = { [KERNEL_HWCAP_ECV] = "ecv", [KERNEL_HWCAP_AFP] = "afp", [KERNEL_HWCAP_RPRES] = "rpres", + [KERNEL_HWCAP_SME] = "sme", + [KERNEL_HWCAP_SME_I16I64] = "smei16i64", + [KERNEL_HWCAP_SME_F64F64] = "smef64f64", + [KERNEL_HWCAP_SME_I8I32] = "smei8i32", + [KERNEL_HWCAP_SME_F16F32] = "smef16f32", + [KERNEL_HWCAP_SME_B16F32] = "smeb16f32", + [KERNEL_HWCAP_SME_F32F32] = "smef32f32", + [KERNEL_HWCAP_SME_FA64] = "smefa64", };

#ifdef CONFIG_COMPAT @@ -400,6 +408,7 @@ static void __cpuinfo_store_cpu(struct cpuinfo_arm64 *info) info->reg_id_aa64pfr0 = read_cpuid(ID_AA64PFR0_EL1); info->reg_id_aa64pfr1 = read_cpuid(ID_AA64PFR1_EL1); info->reg_id_aa64zfr0 = read_cpuid(ID_AA64ZFR0_EL1); + info->reg_id_aa64smfr0 = read_cpuid(ID_AA64SMFR0_EL1);

if (id_aa64pfr1_mte(info->reg_id_aa64pfr1)) info->reg_gmid = read_cpuid(GMID_EL1); diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 5280e098cfb5..576490be3c2b 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -987,6 +987,32 @@ void fpsimd_release_task(struct task_struct *dead_task)

#endif /* CONFIG_ARM64_SVE */

+#ifdef CONFIG_ARM64_SME + +void sme_kernel_enable(const struct arm64_cpu_capabilities *__always_unused p) +{ + /* Set priority for all PEs to architecturally defined minimum */ + write_sysreg_s(read_sysreg_s(SYS_SMPRI_EL1) & ~SMPRI_EL1_PRIORITY_MASK, + SYS_SMPRI_EL1); + + /* Allow SME in kernel */ + write_sysreg(read_sysreg(CPACR_EL1) | CPACR_EL1_SMEN_EL1EN, CPACR_EL1); + isb(); +} + +/* + * This must be called after sme_kernel_enable(), we rely on the + * feature table being sorted to ensure this. + */ +void fa64_kernel_enable(const struct arm64_cpu_capabilities *__always_unused p) +{ + /* Allow use of FA64 */ + write_sysreg_s(read_sysreg_s(SYS_SMCR_EL1) | SMCR_ELx_FA64_MASK, + SYS_SMCR_EL1); +} + +#endif /* CONFIG_ARM64_SVE */ + /* * Trapped SVE access * @@ -1532,6 +1558,10 @@ static int __init fpsimd_init(void) if (!cpu_have_named_feature(ASIMD)) pr_notice("Advanced SIMD is not implemented\n");

+ + if (cpu_have_named_feature(SME) && !cpu_have_named_feature(SVE)) + pr_notice("SME is implemented but not SVE\n"); + return sve_sysctl_init(); } core_initcall(fpsimd_init); diff --git a/arch/arm64/tools/cpucaps b/arch/arm64/tools/cpucaps index 9c65b1e25a96..0a602525bda6 100644 --- a/arch/arm64/tools/cpucaps +++ b/arch/arm64/tools/cpucaps @@ -41,6 +41,8 @@ KVM_PROTECTED_MODE MISMATCHED_CACHE_TYPE MTE MTE_ASYMM +SME +SME_FA64 SPECTRE_V2 SPECTRE_V3A SPECTRE_V4

The vector lengths used for SME are controlled through a similar set of registers to those for SVE and enumerated using a similar algorithm with some slight differences due to the fact that unlike SVE there are no restrictions on which combinations of vector lengths can be supported nor any mandatory vector lengths which must be implemented. Add a new vector type and implement support for enumerating it.

One slightly awkward feature is that we need to read the current vector length using a different instruction (or enter streaming mode which would have the same issue and be higher cost). Rather than add an ops structure we add special cases directly in the otherwise generic vec_probe_vqs() function, this is a bit inelegant but it's the only place where this is an issue.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/cpu.h | 3 + arch/arm64/include/asm/cpufeature.h | 7 ++ arch/arm64/include/asm/fpsimd.h | 26 ++++++ arch/arm64/include/asm/processor.h | 1 + arch/arm64/kernel/cpufeature.c | 47 +++++++++++ arch/arm64/kernel/cpuinfo.c | 4 + arch/arm64/kernel/entry-fpsimd.S | 9 ++ arch/arm64/kernel/fpsimd.c | 123 +++++++++++++++++++++++++++- 8 files changed, 218 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/include/asm/cpu.h b/arch/arm64/include/asm/cpu.h index d08062bcb9c1..550e1fc4ae6c 100644 --- a/arch/arm64/include/asm/cpu.h +++ b/arch/arm64/include/asm/cpu.h @@ -64,6 +64,9 @@ struct cpuinfo_arm64 {

/* pseudo-ZCR for recording maximum ZCR_EL1 LEN value: */ u64 reg_zcr; + + /* pseudo-SMCR for recording maximum ZCR_EL1 LEN value: */ + u64 reg_smcr; };

DECLARE_PER_CPU(struct cpuinfo_arm64, cpu_data); diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h index f93b1442143f..9d36035acce3 100644 --- a/arch/arm64/include/asm/cpufeature.h +++ b/arch/arm64/include/asm/cpufeature.h @@ -620,6 +620,13 @@ static inline bool id_aa64pfr0_sve(u64 pfr0) return val > 0; }

+static inline bool id_aa64pfr1_sme(u64 pfr1) +{ + u32 val = cpuid_feature_extract_unsigned_field(pfr1, ID_AA64PFR1_SME_SHIFT); + + return val > 0; +} + static inline bool id_aa64pfr1_mte(u64 pfr1) { u32 val = cpuid_feature_extract_unsigned_field(pfr1, ID_AA64PFR1_MTE_SHIFT); diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 2e8ef00e7520..32cd682258d9 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -78,6 +78,7 @@ extern void sme_kernel_enable(const struct arm64_cpu_capabilities *__unused); extern void fa64_kernel_enable(const struct arm64_cpu_capabilities *__unused);

extern u64 read_zcr_features(void); +extern u64 read_smcr_features(void);

/* * Helpers to translate bit indices in sve_vq_map to VQ values (and @@ -172,6 +173,12 @@ static inline void write_vl(enum vec_type type, u64 val) tmp = read_sysreg_s(SYS_ZCR_EL1) & ~ZCR_ELx_LEN_MASK; write_sysreg_s(tmp | val, SYS_ZCR_EL1); break; +#endif +#ifdef CONFIG_ARM64_SME + case ARM64_VEC_SME: + tmp = read_sysreg_s(SYS_SMCR_EL1) & ~SMCR_ELx_LEN_MASK; + write_sysreg_s(tmp | val, SYS_SMCR_EL1); + break; #endif default: WARN_ON_ONCE(1); @@ -268,12 +275,31 @@ static inline void sme_smstop(void) asm volatile(__msr_s(SYS_SVCR_SMSTOP_SMZA_EL0, "xzr")); }

+extern void __init sme_setup(void); + +static inline int sme_max_vl(void) +{ + return vec_max_vl(ARM64_VEC_SME); +} + +static inline int sme_max_virtualisable_vl(void) +{ + return vec_max_virtualisable_vl(ARM64_VEC_SME); +} + +extern unsigned int sme_get_vl(void); + #else

static inline void sme_smstart_sm(void) { } static inline void sme_smstop_sm(void) { } static inline void sme_smstop(void) { }

+static inline void sme_setup(void) { } +static inline unsigned int sme_get_vl(void) { return 0; } +static inline int sme_max_vl(void) { return 0; } +static inline int sme_max_virtualisable_vl(void) { return 0; } + #endif /* ! CONFIG_ARM64_SME */

/* For use by EFI runtime services calls only */ diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 6f41b65f9962..df68f0fb0ded 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -117,6 +117,7 @@ struct debug_info {

enum vec_type { ARM64_VEC_SVE = 0, + ARM64_VEC_SME, ARM64_VEC_MAX, };

diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c index 5bfd440edd75..a29347d804b2 100644 --- a/arch/arm64/kernel/cpufeature.c +++ b/arch/arm64/kernel/cpufeature.c @@ -571,6 +571,12 @@ static const struct arm64_ftr_bits ftr_zcr[] = { ARM64_FTR_END, };

+static const struct arm64_ftr_bits ftr_smcr[] = { + ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, + SMCR_ELx_LEN_SHIFT, SMCR_ELx_LEN_SIZE, 0), /* LEN */ + ARM64_FTR_END, +}; + /* * Common ftr bits for a 32bit register with all hidden, strict * attributes, with 4bit feature fields and a default safe value of @@ -674,6 +680,7 @@ static const struct __ftr_reg_entry {

/* Op1 = 0, CRn = 1, CRm = 2 */ ARM64_FTR_REG(SYS_ZCR_EL1, ftr_zcr), + ARM64_FTR_REG(SYS_SMCR_EL1, ftr_smcr),

/* Op1 = 1, CRn = 0, CRm = 0 */ ARM64_FTR_REG(SYS_GMID_EL1, ftr_gmid), @@ -978,6 +985,12 @@ void __init init_cpu_features(struct cpuinfo_arm64 *info) vec_init_vq_map(ARM64_VEC_SVE); }

+ if (id_aa64pfr1_sme(info->reg_id_aa64pfr1)) { + init_cpu_ftr_reg(SYS_SMCR_EL1, info->reg_smcr); + if (IS_ENABLED(CONFIG_ARM64_SME)) + vec_init_vq_map(ARM64_VEC_SME); + } + if (id_aa64pfr1_mte(info->reg_id_aa64pfr1)) init_cpu_ftr_reg(SYS_GMID_EL1, info->reg_gmid);

@@ -1204,6 +1217,9 @@ void update_cpu_features(int cpu, taint |= check_update_ftr_reg(SYS_ID_AA64ZFR0_EL1, cpu, info->reg_id_aa64zfr0, boot->reg_id_aa64zfr0);

+ taint |= check_update_ftr_reg(SYS_ID_AA64SMFR0_EL1, cpu, + info->reg_id_aa64smfr0, boot->reg_id_aa64smfr0); + if (id_aa64pfr0_sve(info->reg_id_aa64pfr0)) { taint |= check_update_ftr_reg(SYS_ZCR_EL1, cpu, info->reg_zcr, boot->reg_zcr); @@ -1214,6 +1230,16 @@ void update_cpu_features(int cpu, vec_update_vq_map(ARM64_VEC_SVE); }

+ if (id_aa64pfr1_sme(info->reg_id_aa64pfr1)) { + taint |= check_update_ftr_reg(SYS_SMCR_EL1, cpu, + info->reg_smcr, boot->reg_smcr); + + /* Probe vector lengths, unless we already gave up on SME */ + if (id_aa64pfr1_sme(read_sanitised_ftr_reg(SYS_ID_AA64PFR1_EL1)) && + !system_capabilities_finalized()) + vec_update_vq_map(ARM64_VEC_SME); + } + /* * The kernel uses the LDGM/STGM instructions and the number of tags * they read/write depends on the GMID_EL1.BS field. Check that the @@ -2883,6 +2909,23 @@ static void verify_sve_features(void) /* Add checks on other ZCR bits here if necessary */ }

+static void verify_sme_features(void) +{ + u64 safe_smcr = read_sanitised_ftr_reg(SYS_SMCR_EL1); + u64 smcr = read_smcr_features(); + + unsigned int safe_len = safe_smcr & SMCR_ELx_LEN_MASK; + unsigned int len = smcr & SMCR_ELx_LEN_MASK; + + if (len < safe_len || vec_verify_vq_map(ARM64_VEC_SME)) { + pr_crit("CPU%d: SME: vector length support mismatch\n", + smp_processor_id()); + cpu_die_early(); + } + + /* Add checks on other SMCR bits here if necessary */ +} + static void verify_hyp_capabilities(void) { u64 safe_mmfr1, mmfr0, mmfr1; @@ -2935,6 +2978,9 @@ static void verify_local_cpu_capabilities(void) if (system_supports_sve()) verify_sve_features();

+ if (system_supports_sme()) + verify_sme_features(); + if (is_hyp_mode_available()) verify_hyp_capabilities(); } @@ -3052,6 +3098,7 @@ void __init setup_cpu_features(void) pr_info("emulated: Privileged Access Never (PAN) using TTBR0_EL1 switching\n");

sve_setup(); + sme_setup(); minsigstksz_setup();

/* Advertise that we have computed the system capabilities */ diff --git a/arch/arm64/kernel/cpuinfo.c b/arch/arm64/kernel/cpuinfo.c index 33ec182e872e..72dc155719b0 100644 --- a/arch/arm64/kernel/cpuinfo.c +++ b/arch/arm64/kernel/cpuinfo.c @@ -420,6 +420,10 @@ static void __cpuinfo_store_cpu(struct cpuinfo_arm64 *info) id_aa64pfr0_sve(info->reg_id_aa64pfr0)) info->reg_zcr = read_zcr_features();

+ if (IS_ENABLED(CONFIG_ARM64_SME) && + id_aa64pfr1_sme(info->reg_id_aa64pfr1)) + info->reg_smcr = read_smcr_features(); + cpuinfo_detect_icache_policy(info); }

diff --git a/arch/arm64/kernel/entry-fpsimd.S b/arch/arm64/kernel/entry-fpsimd.S index dc242e269f9a..deee5f01462e 100644 --- a/arch/arm64/kernel/entry-fpsimd.S +++ b/arch/arm64/kernel/entry-fpsimd.S @@ -86,3 +86,12 @@ SYM_FUNC_START(sve_flush_live) SYM_FUNC_END(sve_flush_live)

#endif /* CONFIG_ARM64_SVE */ + +#ifdef CONFIG_ARM64_SME + +SYM_FUNC_START(sme_get_vl) + _sme_rdsvl 0, 1 + ret +SYM_FUNC_END(sme_get_vl) + +#endif /* CONFIG_ARM64_SME */ diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 576490be3c2b..3fb2167f8af7 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -136,6 +136,12 @@ __ro_after_init struct vl_info vl_info[ARM64_VEC_MAX] = { .max_virtualisable_vl = SVE_VL_MIN, }, #endif +#ifdef CONFIG_ARM64_SME + [ARM64_VEC_SME] = { + .type = ARM64_VEC_SME, + .name = "SME", + }, +#endif };

static unsigned int vec_vl_inherit_flag(enum vec_type type) @@ -186,6 +192,20 @@ extern void __percpu *efi_sve_state;

#endif /* ! CONFIG_ARM64_SVE */

+#ifdef CONFIG_ARM64_SME + +static int get_sme_default_vl(void) +{ + return get_default_vl(ARM64_VEC_SME); +} + +static void set_sme_default_vl(int val) +{ + set_default_vl(ARM64_VEC_SME, val); +} + +#endif + DEFINE_PER_CPU(bool, fpsimd_context_busy); EXPORT_PER_CPU_SYMBOL(fpsimd_context_busy);

@@ -403,6 +423,8 @@ static unsigned int find_supported_vector_length(enum vec_type type,

if (vl > max_vl) vl = max_vl; + if (vl < info->min_vl) + vl = info->min_vl;

bit = find_next_bit(info->vq_map, SVE_VQ_MAX, __vq_to_bit(sve_vq_from_vl(vl))); @@ -764,7 +786,23 @@ static void vec_probe_vqs(struct vl_info *info,

for (vq = SVE_VQ_MAX; vq >= SVE_VQ_MIN; --vq) { write_vl(info->type, vq - 1); /* self-syncing */ - vl = sve_get_vl(); + + switch (info->type) { + case ARM64_VEC_SVE: + vl = sve_get_vl(); + break; + case ARM64_VEC_SME: + vl = sme_get_vl(); + break; + default: + vl = 0; + break; + } + + /* Minimum VL identified? */ + if (sve_vq_from_vl(vl) > vq) + break; + vq = sve_vq_from_vl(vl); /* skip intervening lengths */ set_bit(__vq_to_bit(vq), map); } @@ -1011,7 +1049,88 @@ void fa64_kernel_enable(const struct arm64_cpu_capabilities *__always_unused p) SYS_SMCR_EL1); }

-#endif /* CONFIG_ARM64_SVE */ +/* + * Read the pseudo-SMCR used by cpufeatures to identify the supported + * vector length. + * + * Use only if SME is present. + * This function clobbers the SME vector length. + */ +u64 read_smcr_features(void) +{ + u64 smcr; + unsigned int vq_max; + + sme_kernel_enable(NULL); + sme_smstart_sm(); + + /* + * Set the maximum possible VL. + */ + write_sysreg_s(read_sysreg_s(SYS_SMCR_EL1) | SMCR_ELx_LEN_MASK, + SYS_SMCR_EL1); + + smcr = read_sysreg_s(SYS_SMCR_EL1); + smcr &= ~(u64)SMCR_ELx_LEN_MASK; /* Only the LEN field */ + vq_max = sve_vq_from_vl(sve_get_vl()); + smcr |= vq_max - 1; /* set LEN field to maximum effective value */ + + sme_smstop_sm(); + + return smcr; +} + +void __init sme_setup(void) +{ + struct vl_info *info = &vl_info[ARM64_VEC_SME]; + u64 smcr; + int min_bit; + + if (!system_supports_sme()) + return; + + /* + * SME doesn't require any particular vector length be + * supported but it does require at least one. We should have + * disabled the feature entirely while bringing up CPUs but + * let's double check here. + */ + WARN_ON(bitmap_empty(info->vq_map, SVE_VQ_MAX)); + + min_bit = find_last_bit(info->vq_map, SVE_VQ_MAX); + info->min_vl = sve_vl_from_vq(__bit_to_vq(min_bit)); + + smcr = read_sanitised_ftr_reg(SYS_SMCR_EL1); + info->max_vl = sve_vl_from_vq((smcr & SMCR_ELx_LEN_MASK) + 1); + + /* + * Sanity-check that the max VL we determined through CPU features + * corresponds properly to sme_vq_map. If not, do our best: + */ + if (WARN_ON(info->max_vl != find_supported_vector_length(ARM64_VEC_SME, + info->max_vl))) + info->max_vl = find_supported_vector_length(ARM64_VEC_SME, + info->max_vl); + + WARN_ON(info->min_vl > info->max_vl); + + /* + * For the default VL, pick the maximum supported value <= 32 + * (256 bits) if there is one since this is guaranteed not to + * grow the signal frame when in streaming mode, otherwise the + * minimum available VL will be used. + */ + set_sme_default_vl(find_supported_vector_length(ARM64_VEC_SME, 32)); + + pr_info("SME: minimum available vector length %u bytes per vector\n", + info->min_vl); + pr_info("SME: maximum available vector length %u bytes per vector\n", + info->max_vl); + pr_info("SME: default vector length %u bytes per vector\n", + get_sme_default_vl()); +} + +#endif /* CONFIG_ARM64_SME */

/* * Trapped SVE access

As for SVE provide a prctl() interface which allows processes to configure their SME vector length.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/fpsimd.h | 4 ++++ arch/arm64/include/asm/processor.h | 4 +++- arch/arm64/include/asm/thread_info.h | 1 + arch/arm64/kernel/fpsimd.c | 32 ++++++++++++++++++++++++++++ include/uapi/linux/prctl.h | 9 ++++++++ kernel/sys.c | 12 +++++++++++ 6 files changed, 61 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 32cd682258d9..38fd6aab7feb 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -288,6 +288,8 @@ static inline int sme_max_virtualisable_vl(void) }

extern unsigned int sme_get_vl(void); +extern int sme_set_current_vl(unsigned long arg); +extern int sme_get_current_vl(void);

#else

@@ -299,6 +301,8 @@ static inline void sme_setup(void) { } static inline unsigned int sme_get_vl(void) { return 0; } static inline int sme_max_vl(void) { return 0; } static inline int sme_max_virtualisable_vl(void) { return 0; } +static inline int sme_set_current_vl(unsigned long arg) { return -EINVAL; } +static inline int sme_get_current_vl(void) { return -EINVAL; }

#endif /* ! CONFIG_ARM64_SME */

diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index df68f0fb0ded..f2c2ebd440e2 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -354,9 +354,11 @@ extern void __init minsigstksz_setup(void); */ #include <asm/fpsimd.h>

-/* Userspace interface for PR_SVE_{SET,GET}_VL prctl()s: */ +/* Userspace interface for PR_S[MV]E_{SET,GET}_VL prctl()s: */ #define SVE_SET_VL(arg) sve_set_current_vl(arg) #define SVE_GET_VL() sve_get_current_vl() +#define SME_SET_VL(arg) sme_set_current_vl(arg) +#define SME_GET_VL() sme_get_current_vl()

/* PR_PAC_RESET_KEYS prctl */ #define PAC_RESET_KEYS(tsk, arg) ptrauth_prctl_reset_keys(tsk, arg) diff --git a/arch/arm64/include/asm/thread_info.h b/arch/arm64/include/asm/thread_info.h index e1317b7c4525..4e6b58dcd6f9 100644 --- a/arch/arm64/include/asm/thread_info.h +++ b/arch/arm64/include/asm/thread_info.h @@ -82,6 +82,7 @@ int arch_dup_task_struct(struct task_struct *dst, #define TIF_SVE_VL_INHERIT 24 /* Inherit SVE vl_onexec across exec */ #define TIF_SSBD 25 /* Wants SSB mitigation */ #define TIF_TAGGED_ADDR 26 /* Allow tagged user addresses */ +#define TIF_SME_VL_INHERIT 28 /* Inherit SME vl_onexec across exec */

#define _TIF_SIGPENDING (1 << TIF_SIGPENDING) #define _TIF_NEED_RESCHED (1 << TIF_NEED_RESCHED) diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 841b9609342f..1edb3996f9cf 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -149,6 +149,8 @@ static unsigned int vec_vl_inherit_flag(enum vec_type type) switch (type) { case ARM64_VEC_SVE: return TIF_SVE_VL_INHERIT; + case ARM64_VEC_SME: + return TIF_SME_VL_INHERIT; default: WARN_ON_ONCE(1); return 0; @@ -801,6 +803,36 @@ int sve_get_current_vl(void) return vec_prctl_status(ARM64_VEC_SVE, 0); }

+#ifdef CONFIG_ARM64_SME +/* PR_SME_SET_VL */ +int sme_set_current_vl(unsigned long arg) +{ + unsigned long vl, flags; + int ret; + + vl = arg & PR_SME_VL_LEN_MASK; + flags = arg & ~vl; + + if (!system_supports_sme() || is_compat_task()) + return -EINVAL; + + ret = vec_set_vector_length(current, ARM64_VEC_SME, vl, flags); + if (ret) + return ret; + + return vec_prctl_status(ARM64_VEC_SME, flags); +} + +/* PR_SME_GET_VL */ +int sme_get_current_vl(void) +{ + if (!system_supports_sme() || is_compat_task()) + return -EINVAL; + + return vec_prctl_status(ARM64_VEC_SME, 0); +} +#endif /* CONFIG_ARM64_SME */ + static void vec_probe_vqs(struct vl_info *info, DECLARE_BITMAP(map, SVE_VQ_MAX)) { diff --git a/include/uapi/linux/prctl.h b/include/uapi/linux/prctl.h index e998764f0262..a5e06dcbba13 100644 --- a/include/uapi/linux/prctl.h +++ b/include/uapi/linux/prctl.h @@ -272,6 +272,15 @@ struct prctl_mm_map { # define PR_SCHED_CORE_SCOPE_THREAD_GROUP 1 # define PR_SCHED_CORE_SCOPE_PROCESS_GROUP 2

+/* arm64 Scalable Matrix Extension controls */ +/* Flag values must be in sync with SVE versions */ +#define PR_SME_SET_VL 63 /* set task vector length */ +# define PR_SME_SET_VL_ONEXEC (1 << 18) /* defer effect until exec */ +#define PR_SME_GET_VL 64 /* get task vector length */ +/* Bits common to PR_SME_SET_VL and PR_SME_GET_VL */ +# define PR_SME_VL_LEN_MASK 0xffff +# define PR_SME_VL_INHERIT (1 << 17) /* inherit across exec */ + #define PR_SET_VMA 0x53564d41 # define PR_SET_VMA_ANON_NAME 0

diff --git a/kernel/sys.c b/kernel/sys.c index ecc4cf019242..5cd4fffc7cd0 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -116,6 +116,12 @@ #ifndef SVE_GET_VL # define SVE_GET_VL() (-EINVAL) #endif +#ifndef SME_SET_VL +# define SME_SET_VL(a) (-EINVAL) +#endif +#ifndef SME_GET_VL +# define SME_GET_VL() (-EINVAL) +#endif #ifndef PAC_RESET_KEYS # define PAC_RESET_KEYS(a, b) (-EINVAL) #endif @@ -2523,6 +2529,12 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3, case PR_SVE_GET_VL: error = SVE_GET_VL(); break; + case PR_SME_SET_VL: + error = SME_SET_VL(arg2); + break; + case PR_SME_GET_VL: + error = SME_GET_VL(); + break; case PR_GET_SPECULATION_CTRL: if (arg3 || arg4 || arg5) return -EINVAL;

In SME the use of both streaming SVE mode and ZA are tracked through PSTATE.SM and PSTATE.ZA, visible through the system register SVCR. In order to context switch the floating point state for SME we need to context switch the contents of this register as part of context switching the floating point state.

Since changing the vector length exits streaming SVE mode and disables ZA we also make sure we update SVCR appropriately when setting vector length, and similarly ensure that new threads have streaming SVE mode and ZA disabled.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/fpsimd.h | 3 ++- arch/arm64/include/asm/processor.h | 1 + arch/arm64/include/asm/thread_info.h | 1 + arch/arm64/kernel/fpsimd.c | 18 +++++++++++++++++- arch/arm64/kernel/process.c | 2 ++ arch/arm64/kvm/fpsimd.c | 4 ++++ 6 files changed, 27 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index 38fd6aab7feb..821d270980da 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -46,7 +46,8 @@ extern void fpsimd_restore_current_state(void); extern void fpsimd_update_current_state(struct user_fpsimd_state const *state);

extern void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *state, - void *sve_state, unsigned int sve_vl); + void *sve_state, unsigned int sve_vl, + u64 *svcr);

extern void fpsimd_flush_task_state(struct task_struct *target); extern void fpsimd_save_and_flush_cpu_state(void); diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 008a1767ebff..7e08a4d48c24 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -168,6 +168,7 @@ struct thread_struct { u64 mte_ctrl; #endif u64 sctlr_user; + u64 svcr; u64 tpidr2_el0; };

diff --git a/arch/arm64/include/asm/thread_info.h b/arch/arm64/include/asm/thread_info.h index 4e6b58dcd6f9..848739c15de8 100644 --- a/arch/arm64/include/asm/thread_info.h +++ b/arch/arm64/include/asm/thread_info.h @@ -82,6 +82,7 @@ int arch_dup_task_struct(struct task_struct *dst, #define TIF_SVE_VL_INHERIT 24 /* Inherit SVE vl_onexec across exec */ #define TIF_SSBD 25 /* Wants SSB mitigation */ #define TIF_TAGGED_ADDR 26 /* Allow tagged user addresses */ +#define TIF_SME 27 /* SME in use */ #define TIF_SME_VL_INHERIT 28 /* Inherit SME vl_onexec across exec */

#define _TIF_SIGPENDING (1 << TIF_SIGPENDING) diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 40ef89120774..a1918b71d335 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -121,6 +121,7 @@ struct fpsimd_last_state_struct { struct user_fpsimd_state *st; void *sve_state; + u64 *svcr; unsigned int sve_vl; };

@@ -359,6 +360,9 @@ static void task_fpsimd_load(void) WARN_ON(!system_supports_fpsimd()); WARN_ON(!have_cpu_fpsimd_context());

+ if (IS_ENABLED(CONFIG_ARM64_SME) && test_thread_flag(TIF_SME)) + write_sysreg_s(current->thread.svcr, SYS_SVCR_EL0); + if (IS_ENABLED(CONFIG_ARM64_SVE) && test_thread_flag(TIF_SVE)) { sve_set_vq(sve_vq_from_vl(task_get_sve_vl(current)) - 1); sve_load_state(sve_pffr(&current->thread), @@ -384,6 +388,12 @@ static void fpsimd_save(void) if (test_thread_flag(TIF_FOREIGN_FPSTATE)) return;

+ if (IS_ENABLED(CONFIG_ARM64_SME) && + test_thread_flag(TIF_SME)) { + u64 *svcr = last->svcr; + *svcr = read_sysreg_s(SYS_SVCR_EL0); + } + if (IS_ENABLED(CONFIG_ARM64_SVE) && test_thread_flag(TIF_SVE)) { if (WARN_ON(sve_get_vl() != last->sve_vl)) { @@ -735,6 +745,10 @@ int vec_set_vector_length(struct task_struct *task, enum vec_type type, if (test_and_clear_tsk_thread_flag(task, TIF_SVE)) sve_to_fpsimd(task);

+ if (system_supports_sme() && type == ARM64_VEC_SME) + task->thread.svcr &= ~(SYS_SVCR_EL0_SM_MASK | + SYS_SVCR_EL0_ZA_MASK); + if (task == current) put_cpu_fpsimd_context();

@@ -1398,6 +1412,7 @@ static void fpsimd_bind_task_to_cpu(void) last->st = &current->thread.uw.fpsimd_state; last->sve_state = current->thread.sve_state; last->sve_vl = task_get_sve_vl(current); + last->svcr = &current->thread.svcr; current->thread.fpsimd_cpu = smp_processor_id();

if (system_supports_sve()) { @@ -1412,7 +1427,7 @@ static void fpsimd_bind_task_to_cpu(void) }

void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, - unsigned int sve_vl) + unsigned int sve_vl, u64 *svcr) { struct fpsimd_last_state_struct *last = this_cpu_ptr(&fpsimd_last_state); @@ -1421,6 +1436,7 @@ void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, WARN_ON(!in_softirq() && !irqs_disabled());

last->st = st; + last->svcr = svcr; last->sve_state = sve_state; last->sve_vl = sve_vl; } diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c index e69a3dcdb0d9..f2d32a29641c 100644 --- a/arch/arm64/kernel/process.c +++ b/arch/arm64/kernel/process.c @@ -310,6 +310,8 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src) dst->thread.sve_state = NULL; clear_tsk_thread_flag(dst, TIF_SVE);

+ dst->thread.svcr = 0; + /* clear any pending asynchronous tag fault raised by the parent */ clear_tsk_thread_flag(dst, TIF_MTE_ASYNC_FAULT);

diff --git a/arch/arm64/kvm/fpsimd.c b/arch/arm64/kvm/fpsimd.c index 2f48fd362a8c..04698c4bcd30 100644 --- a/arch/arm64/kvm/fpsimd.c +++ b/arch/arm64/kvm/fpsimd.c @@ -103,6 +103,10 @@ void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu) WARN_ON_ONCE(!irqs_disabled());

if (vcpu->arch.flags & KVM_ARM64_FP_ENABLED) { + /* + * Currently we do not support SME guests so SVCR is + * always 0 and we just need a variable to point to. + */ fpsimd_bind_state_to_cpu(&vcpu->arch.ctxt.fp_regs, vcpu->arch.sve_state, vcpu->arch.sve_max_vl);

Allocate space for storing ZA on first access to SME and use that to save and restore ZA state when context switching. We do this by using the vector form of the LDR and STR ZA instructions, these do not require streaming mode and have implementation recommendations that they avoid contention issues in shared SMCU implementations.

Since ZA is architecturally guaranteed to be zeroed when enabled we do not need to explicitly zero ZA, either we will be restoring from a saved copy or trapping on first use of SME so we know that ZA must be disabled.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/fpsimd.h | 5 ++++- arch/arm64/include/asm/fpsimdmacros.h | 22 ++++++++++++++++++++++ arch/arm64/include/asm/kvm_host.h | 3 +++ arch/arm64/include/asm/processor.h | 1 + arch/arm64/kernel/entry-fpsimd.S | 22 ++++++++++++++++++++++ arch/arm64/kernel/fpsimd.c | 20 +++++++++++++------- arch/arm64/kvm/fpsimd.c | 2 +- 7 files changed, 66 insertions(+), 9 deletions(-)

diff --git a/arch/arm64/include/asm/fpsimd.h b/arch/arm64/include/asm/fpsimd.h index cd94f5c5b516..1a709c03bb6c 100644 --- a/arch/arm64/include/asm/fpsimd.h +++ b/arch/arm64/include/asm/fpsimd.h @@ -47,7 +47,8 @@ extern void fpsimd_update_current_state(struct user_fpsimd_state const *state);

extern void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *state, void *sve_state, unsigned int sve_vl, - unsigned int sme_vl, u64 *svcr); + void *za_state, unsigned int sme_vl, + u64 *svcr);

extern void fpsimd_flush_task_state(struct task_struct *target); extern void fpsimd_save_and_flush_cpu_state(void); @@ -90,6 +91,8 @@ extern void sve_flush_live(bool flush_ffr, unsigned long vq_minus_1); extern unsigned int sve_get_vl(void); extern void sve_set_vq(unsigned long vq_minus_1); extern void sme_set_vq(unsigned long vq_minus_1); +extern void za_save_state(void *state); +extern void za_load_state(void const *state);

struct arm64_cpu_capabilities; extern void sve_kernel_enable(const struct arm64_cpu_capabilities *__unused); diff --git a/arch/arm64/include/asm/fpsimdmacros.h b/arch/arm64/include/asm/fpsimdmacros.h index f6ab36e0cd8d..5e0910cf4832 100644 --- a/arch/arm64/include/asm/fpsimdmacros.h +++ b/arch/arm64/include/asm/fpsimdmacros.h @@ -319,3 +319,25 @@ ldr w\nxtmp, [\xpfpsr, #4] msr fpcr, x\nxtmp .endm + +.macro sme_save_za nxbase, xvl, nw + mov w\nw, #0 + +423: + _sme_str_zav \nw, \nxbase + add x\nxbase, x\nxbase, \xvl + add x\nw, x\nw, #1 + cmp \xvl, x\nw + bne 423b +.endm + +.macro sme_load_za nxbase, xvl, nw + mov w\nw, #0 + +423: + _sme_ldr_zav \nw, \nxbase + add x\nxbase, x\nxbase, \xvl + add x\nw, x\nw, #1 + cmp \xvl, x\nw + bne 423b +.endm diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h index 5bc01e62c08a..7dc85d5a6552 100644 --- a/arch/arm64/include/asm/kvm_host.h +++ b/arch/arm64/include/asm/kvm_host.h @@ -280,8 +280,11 @@ struct vcpu_reset_state {

struct kvm_vcpu_arch { struct kvm_cpu_context ctxt; + + /* Guest floating point state */ void *sve_state; unsigned int sve_max_vl; + u64 svcr;

/* Stage 2 paging state used by the hardware on next switch */ struct kvm_s2_mmu *hw_mmu; diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 6e2af9de153c..5a5c5edd76df 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -153,6 +153,7 @@ struct thread_struct {

unsigned int fpsimd_cpu; void *sve_state; /* SVE registers, if any */ + void *za_state; /* ZA register, if any */ unsigned int vl[ARM64_VEC_MAX]; /* vector length */ unsigned int vl_onexec[ARM64_VEC_MAX]; /* vl after next exec */ unsigned long fault_address; /* fault info */ diff --git a/arch/arm64/kernel/entry-fpsimd.S b/arch/arm64/kernel/entry-fpsimd.S index 6f88c0f86d50..229436f33df5 100644 --- a/arch/arm64/kernel/entry-fpsimd.S +++ b/arch/arm64/kernel/entry-fpsimd.S @@ -99,4 +99,26 @@ SYM_FUNC_START(sme_set_vq) ret SYM_FUNC_END(sme_set_vq)

+/* + * Save the SME state + * + * x0 - pointer to buffer for state + */ +SYM_FUNC_START(za_save_state) + _sme_rdsvl 1, 1 // x1 = VL/8 + sme_save_za 0, x1, 12 + ret +SYM_FUNC_END(za_save_state) + +/* + * Load the SME state + * + * x0 - pointer to buffer for state + */ +SYM_FUNC_START(za_load_state) + _sme_rdsvl 1, 1 // x1 = VL/8 + sme_load_za 0, x1, 12 + ret +SYM_FUNC_END(za_load_state) + #endif /* CONFIG_ARM64_SME */ diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 12fef62cf07a..c9e8186e69c0 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -121,6 +121,7 @@ struct fpsimd_last_state_struct { struct user_fpsimd_state *st; void *sve_state; + void *za_state; u64 *svcr; unsigned int sve_vl; unsigned int sme_vl; @@ -387,11 +388,15 @@ static void task_fpsimd_load(void) if (system_supports_sme()) { unsigned long sme_vl = task_get_sme_vl(current);

+ /* Ensure VL is set up for restoring data */ if (test_thread_flag(TIF_SME)) sme_set_vq(sve_vq_from_vl(sme_vl) - 1);

write_sysreg_s(current->thread.svcr, SYS_SVCR_EL0);

+ if (thread_za_enabled(&current->thread)) + za_load_state(current->thread.za_state); + if (thread_sm_enabled(&current->thread)) { restore_sve_regs = true; restore_ffr = system_supports_fa64(); @@ -435,11 +440,10 @@ static void fpsimd_save(void) u64 *svcr = last->svcr; *svcr = read_sysreg_s(SYS_SVCR_EL0);

- if (thread_za_enabled(&current->thread)) { - /* ZA state managment is not implemented yet */ - force_signal_inject(SIGKILL, SI_KERNEL, 0, 0); - return; - } + *svcr = read_sysreg_s(SYS_SVCR_EL0); + + if (*svcr & SYS_SVCR_EL0_ZA_MASK) + za_save_state(last->za_state);

/* If we are in streaming mode override regular SVE. */ if (*svcr & SYS_SVCR_EL0_SM_MASK) { @@ -1477,6 +1481,7 @@ static void fpsimd_bind_task_to_cpu(void) WARN_ON(!system_supports_fpsimd()); last->st = &current->thread.uw.fpsimd_state; last->sve_state = current->thread.sve_state; + last->za_state = current->thread.za_state; last->sve_vl = task_get_sve_vl(current); last->sme_vl = task_get_sme_vl(current); last->svcr = &current->thread.svcr; @@ -1494,8 +1499,8 @@ static void fpsimd_bind_task_to_cpu(void) }

void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, - unsigned int sve_vl, unsigned int sme_vl, - u64 *svcr) + unsigned int sve_vl, void *za_state, + unsigned int sme_vl, u64 *svcr) { struct fpsimd_last_state_struct *last = this_cpu_ptr(&fpsimd_last_state); @@ -1506,6 +1511,7 @@ void fpsimd_bind_state_to_cpu(struct user_fpsimd_state *st, void *sve_state, last->st = st; last->svcr = svcr; last->sve_state = sve_state; + last->za_state = za_state; last->sve_vl = sve_vl; last->sme_vl = sme_vl; } diff --git a/arch/arm64/kvm/fpsimd.c b/arch/arm64/kvm/fpsimd.c index 902c598b7ed2..338733ac63f8 100644 --- a/arch/arm64/kvm/fpsimd.c +++ b/arch/arm64/kvm/fpsimd.c @@ -110,7 +110,7 @@ void kvm_arch_vcpu_ctxsync_fp(struct kvm_vcpu *vcpu) fpsimd_bind_state_to_cpu(&vcpu->arch.ctxt.fp_regs, vcpu->arch.sve_state, vcpu->arch.sve_max_vl, - 0); + NULL, 0, &vcpu->arch.svcr);

clear_thread_flag(TIF_FOREIGN_FPSTATE); update_thread_flag(TIF_SVE, vcpu_has_sve(vcpu));

When in streaming mode we have the same set of SVE registers as we do in regular SVE mode with the exception of FFR and the use of the SME vector length. Provide signal handling for these registers by taking one of the reserved words in the SVE signal context as a flags field and defining a flag which is set for streaming mode. When the flag is set the vector length is set to the streaming mode vector length and we save and restore streaming mode data. We support entering or leaving streaming mode based on the value of the flag but do not support changing the vector length, this is not currently supported SVE signal handling.

We could instead allocate a separate record in the signal frame for the streaming mode SVE context but this inflates the size of the maximal signal frame required and adds complication when validating signal frames from userspace, especially given the current structure of the code.

Any implementation of support for streaming mode vectors in signals will have some potential for causing issues for applications that attempt to handle SVE vectors in signals, use streaming mode but do not understand streaming mode in their signal handling code, it is hard to identify a case that is clearly better than any other - they all have cases where they could cause unexpected register corruption or faults.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/asm/processor.h | 8 +++++ arch/arm64/include/uapi/asm/sigcontext.h | 16 +++++++-- arch/arm64/kernel/signal.c | 42 ++++++++++++++++++------ 3 files changed, 53 insertions(+), 13 deletions(-)

diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h index 5a5c5edd76df..29e0f87b5dd5 100644 --- a/arch/arm64/include/asm/processor.h +++ b/arch/arm64/include/asm/processor.h @@ -189,6 +189,14 @@ static inline unsigned int thread_get_sme_vl(struct thread_struct *thread) return thread_get_vl(thread, ARM64_VEC_SME); }

+static inline unsigned int thread_get_cur_vl(struct thread_struct *thread) +{ + if (system_supports_sme() && (thread->svcr & SYS_SVCR_EL0_SM_MASK)) + return thread_get_sme_vl(thread); + else + return thread_get_sve_vl(thread); +} + unsigned int task_get_vl(const struct task_struct *task, enum vec_type type); void task_set_vl(struct task_struct *task, enum vec_type type, unsigned long vl); diff --git a/arch/arm64/include/uapi/asm/sigcontext.h b/arch/arm64/include/uapi/asm/sigcontext.h index 0c796c795dbe..3a3366d4fbc2 100644 --- a/arch/arm64/include/uapi/asm/sigcontext.h +++ b/arch/arm64/include/uapi/asm/sigcontext.h @@ -134,9 +134,12 @@ struct extra_context { struct sve_context { struct _aarch64_ctx head; __u16 vl; - __u16 __reserved[3]; + __u16 flags; + __u16 __reserved[2]; };

+#define SVE_SIG_FLAG_SM 0x1 /* Context describes streaming mode */ + #endif /* !__ASSEMBLY__ */

#include <asm/sve_context.h> @@ -186,9 +189,16 @@ struct sve_context { * sve_context.vl must equal the thread's current vector length when * doing a sigreturn. * + * On systems with support for SME the SVE register state may reflect either + * streaming or non-streaming mode. In streaming mode the streaming mode + * vector length will be used and the flag SVE_SIG_FLAG_SM will be set in + * the flags field. It is permitted to enter or leave streaming mode in + * a signal return, applications should take care to ensure that any difference + * in vector length between the two modes is handled, including any resixing + * and movement of context blocks. * - * Note: for all these macros, the "vq" argument denotes the SVE - * vector length in quadwords (i.e., units of 128 bits). + * Note: for all these macros, the "vq" argument denotes the vector length + * in quadwords (i.e., units of 128 bits). * * The correct way to obtain vq is to use sve_vq_from_vl(vl). The * result is valid if and only if sve_vl_valid(vl) is true. This is diff --git a/arch/arm64/kernel/signal.c b/arch/arm64/kernel/signal.c index cda04fd73333..8931954e7b39 100644 --- a/arch/arm64/kernel/signal.c +++ b/arch/arm64/kernel/signal.c @@ -227,11 +227,17 @@ static int preserve_sve_context(struct sve_context __user *ctx) { int err = 0; u16 reserved[ARRAY_SIZE(ctx->__reserved)]; + u16 flags = 0; unsigned int vl = task_get_sve_vl(current); unsigned int vq = 0;

- if (test_thread_flag(TIF_SVE)) + if (thread_sm_enabled(&current->thread)) { + vl = task_get_sme_vl(current); vq = sve_vq_from_vl(vl); + flags |= SVE_SIG_FLAG_SM; + } else if (test_thread_flag(TIF_SVE)) { + vq = sve_vq_from_vl(vl); + }

memset(reserved, 0, sizeof(reserved));

@@ -239,6 +245,7 @@ static int preserve_sve_context(struct sve_context __user *ctx) __put_user_error(round_up(SVE_SIG_CONTEXT_SIZE(vq), 16), &ctx->head.size, err); __put_user_error(vl, &ctx->vl, err); + __put_user_error(flags, &ctx->flags, err); BUILD_BUG_ON(sizeof(ctx->__reserved) != sizeof(reserved)); err |= __copy_to_user(&ctx->__reserved, reserved, sizeof(reserved));

@@ -259,18 +266,28 @@ static int preserve_sve_context(struct sve_context __user *ctx) static int restore_sve_fpsimd_context(struct user_ctxs *user) { int err; - unsigned int vq; + unsigned int vl, vq; struct user_fpsimd_state fpsimd; struct sve_context sve;

if (__copy_from_user(&sve, user->sve, sizeof(sve))) return -EFAULT;

- if (sve.vl != task_get_sve_vl(current)) + if (sve.flags & SVE_SIG_FLAG_SM) { + if (!system_supports_sme()) + return -EINVAL; + + vl = task_get_sme_vl(current); + } else { + vl = task_get_sve_vl(current); + } + + if (sve.vl != vl) return -EINVAL;

if (sve.head.size <= sizeof(*user->sve)) { clear_thread_flag(TIF_SVE); + current->thread.svcr &= ~SYS_SVCR_EL0_SM_MASK; goto fpsimd_only; }

@@ -302,7 +319,10 @@ static int restore_sve_fpsimd_context(struct user_ctxs *user) if (err) return -EFAULT;

- set_thread_flag(TIF_SVE); + if (sve.flags & SVE_SIG_FLAG_SM) + current->thread.svcr |= SYS_SVCR_EL0_SM_MASK; + else + set_thread_flag(TIF_SVE);

fpsimd_only: /* copy the FP and status/control registers */ @@ -394,7 +414,7 @@ static int parse_user_sigframe(struct user_ctxs *user, break;

case SVE_MAGIC: - if (!system_supports_sve()) + if (!system_supports_sve() && !system_supports_sme()) goto invalid;

if (user->sve) @@ -593,11 +613,12 @@ static int setup_sigframe_layout(struct rt_sigframe_user_layout *user, if (system_supports_sve()) { unsigned int vq = 0;

- if (add_all || test_thread_flag(TIF_SVE)) { - int vl = sve_max_vl(); + if (add_all || test_thread_flag(TIF_SVE) || + thread_sm_enabled(&current->thread)) { + int vl = max(sve_max_vl(), sme_max_vl());

if (!add_all) - vl = task_get_sve_vl(current); + vl = thread_get_cur_vl(&current->thread);

vq = sve_vq_from_vl(vl); } @@ -648,8 +669,9 @@ static int setup_sigframe(struct rt_sigframe_user_layout *user, __put_user_error(current->thread.fault_code, &esr_ctx->esr, err); }

- /* Scalable Vector Extension state, if present */ - if (system_supports_sve() && err == 0 && user->sve_offset) { + /* Scalable Vector Extension state (including streaming), if present */ + if ((system_supports_sve() || system_supports_sme()) && + err == 0 && user->sve_offset) { struct sve_context __user *sve_ctx = apply_user_offset(user, user->sve_offset); err |= preserve_sve_context(sve_ctx);

The streaming mode SVE registers are represented using the same data structures as for SVE but since the vector lengths supported and in use may not be the same as SVE we represent them with a new type NT_ARM_SSVE. Unfortunately we only have a single 16 bit reserved field available in the header so there is no space to fit the current and maximum vector length for both standard and streaming SVE mode without redefining the structure in a way the creates a complicatd and fragile ABI. Since FFR is not present in streaming mode it is read and written as zero.

Setting NT_ARM_SSVE registers will put the task into streaming mode, similarly setting NT_ARM_SVE registers will exit it. Reads that do not correspond to the current mode of the task will return the header with no register data. For compatibility reasons on write setting no flag for the register type will be interpreted as setting SVE registers, though users can provide no register data as an alternative mechanism for doing so.

Signed-off-by: Mark Brown broonie@kernel.org --- arch/arm64/include/uapi/asm/ptrace.h | 13 +- arch/arm64/kernel/fpsimd.c | 14 +- arch/arm64/kernel/ptrace.c | 212 +++++++++++++++++++++------ include/uapi/linux/elf.h | 1 + 4 files changed, 184 insertions(+), 56 deletions(-)

diff --git a/arch/arm64/include/uapi/asm/ptrace.h b/arch/arm64/include/uapi/asm/ptrace.h index 758ae984ff97..522b925a78c1 100644 --- a/arch/arm64/include/uapi/asm/ptrace.h +++ b/arch/arm64/include/uapi/asm/ptrace.h @@ -109,7 +109,7 @@ struct user_hwdebug_state { } dbg_regs[16]; };

-/* SVE/FP/SIMD state (NT_ARM_SVE) */ +/* SVE/FP/SIMD state (NT_ARM_SVE & NT_ARM_SSVE) */

struct user_sve_header { __u32 size; /* total meaningful regset content in bytes */ @@ -220,6 +220,7 @@ struct user_sve_header { (SVE_PT_SVE_PREG_OFFSET(vq, __SVE_NUM_PREGS) - \ SVE_PT_SVE_PREGS_OFFSET(vq))

+/* For streaming mode SVE (SSVE) FFR must be read and written as zero */ #define SVE_PT_SVE_FFR_OFFSET(vq) \ (SVE_PT_REGS_OFFSET + __SVE_FFR_OFFSET(vq))

@@ -240,10 +241,12 @@ struct user_sve_header { - SVE_PT_SVE_OFFSET + (__SVE_VQ_BYTES - 1)) \ / __SVE_VQ_BYTES * __SVE_VQ_BYTES)

-#define SVE_PT_SIZE(vq, flags) \ - (((flags) & SVE_PT_REGS_MASK) == SVE_PT_REGS_SVE ? \ - SVE_PT_SVE_OFFSET + SVE_PT_SVE_SIZE(vq, flags) \ - : SVE_PT_FPSIMD_OFFSET + SVE_PT_FPSIMD_SIZE(vq, flags)) +#define SVE_PT_SIZE(vq, flags) \ + (((flags) & SVE_PT_REGS_MASK) == SVE_PT_REGS_SVE ? \ + SVE_PT_SVE_OFFSET + SVE_PT_SVE_SIZE(vq, flags) \ + : ((((flags) & SVE_PT_REGS_MASK) == SVE_PT_REGS_FPSIMD ? \ + SVE_PT_FPSIMD_OFFSET + SVE_PT_FPSIMD_SIZE(vq, flags) \ + : SVE_PT_REGS_OFFSET)))

/* pointer authentication masks (NT_ARM_PAC_MASK) */

diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 487b616ce380..2e931373b361 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -637,7 +637,7 @@ static void fpsimd_to_sve(struct task_struct *task) if (!system_supports_sve()) return;

- vq = sve_vq_from_vl(task_get_sve_vl(task)); + vq = sve_vq_from_vl(thread_get_cur_vl(&task->thread)); __fpsimd_to_sve(sst, fst, vq); }

@@ -654,7 +654,7 @@ static void fpsimd_to_sve(struct task_struct *task) */ static void sve_to_fpsimd(struct task_struct *task) { - unsigned int vq; + unsigned int vq, vl; void const *sst = task->thread.sve_state; struct user_fpsimd_state *fst = &task->thread.uw.fpsimd_state; unsigned int i; @@ -663,7 +663,12 @@ static void sve_to_fpsimd(struct task_struct *task) if (!system_supports_sve()) return;

- vq = sve_vq_from_vl(task_get_sve_vl(task)); + if (thread_sm_enabled(&task->thread)) + vl = task_get_sme_vl(task); + else + vl = task_get_sve_vl(task); + + vq = sve_vq_from_vl(vl); for (i = 0; i < SVE_NUM_ZREGS; ++i) { p = (__uint128_t const *)ZREG(sst, vq, i); fst->vregs[i] = arm64_le128_to_cpu(*p); @@ -804,8 +809,7 @@ int vec_set_vector_length(struct task_struct *task, enum vec_type type, /* * To ensure the FPSIMD bits of the SVE vector registers are preserved, * write any live register state back to task_struct, and convert to a - * regular FPSIMD thread. Since the vector length can only be changed - * with a syscall we can't be in streaming mode while reconfiguring. + * regular FPSIMD thread. */ if (task == current) { get_cpu_fpsimd_context(); diff --git a/arch/arm64/kernel/ptrace.c b/arch/arm64/kernel/ptrace.c index 39dbdfdc38d3..0b8324fa1abc 100644 --- a/arch/arm64/kernel/ptrace.c +++ b/arch/arm64/kernel/ptrace.c @@ -714,21 +714,51 @@ static int system_call_set(struct task_struct *target, #ifdef CONFIG_ARM64_SVE

static void sve_init_header_from_task(struct user_sve_header *header, - struct task_struct *target) + struct task_struct *target, + enum vec_type type) { unsigned int vq; + bool active; + bool fpsimd_only; + enum vec_type task_type;

memset(header, 0, sizeof(*header));

- header->flags = test_tsk_thread_flag(target, TIF_SVE) ? - SVE_PT_REGS_SVE : SVE_PT_REGS_FPSIMD; - if (test_tsk_thread_flag(target, TIF_SVE_VL_INHERIT)) - header->flags |= SVE_PT_VL_INHERIT; + /* Check if the requested registers are active for the task */ + if (thread_sm_enabled(&target->thread)) + task_type = ARM64_VEC_SME; + else + task_type = ARM64_VEC_SVE; + active = (task_type == type); + + switch (type) { + case ARM64_VEC_SVE: + if (test_tsk_thread_flag(target, TIF_SVE_VL_INHERIT)) + header->flags |= SVE_PT_VL_INHERIT; + fpsimd_only = !test_tsk_thread_flag(target, TIF_SVE); + break; + case ARM64_VEC_SME: + if (test_tsk_thread_flag(target, TIF_SME_VL_INHERIT)) + header->flags |= SVE_PT_VL_INHERIT; + fpsimd_only = false; + break; + default: + WARN_ON_ONCE(1); + return; + }

- header->vl = task_get_sve_vl(target); + if (active) { + if (fpsimd_only) { + header->flags |= SVE_PT_REGS_FPSIMD; + } else { + header->flags |= SVE_PT_REGS_SVE; + } + } + + header->vl = task_get_vl(target, type); vq = sve_vq_from_vl(header->vl);

- header->max_vl = sve_max_vl(); + header->max_vl = vec_max_vl(type); header->size = SVE_PT_SIZE(vq, header->flags); header->max_size = SVE_PT_SIZE(sve_vq_from_vl(header->max_vl), SVE_PT_REGS_SVE); @@ -739,19 +769,17 @@ static unsigned int sve_size_from_header(struct user_sve_header const *header) return ALIGN(header->size, SVE_VQ_BYTES); }

-static int sve_get(struct task_struct *target, - const struct user_regset *regset, - struct membuf to) +static int sve_get_common(struct task_struct *target, + const struct user_regset *regset, + struct membuf to, + enum vec_type type) { struct user_sve_header header; unsigned int vq; unsigned long start, end;

- if (!system_supports_sve()) - return -EINVAL; - /* Header */ - sve_init_header_from_task(&header, target); + sve_init_header_from_task(&header, target, type); vq = sve_vq_from_vl(header.vl);

membuf_write(&to, &header, sizeof(header)); @@ -759,49 +787,61 @@ static int sve_get(struct task_struct *target, if (target == current) fpsimd_preserve_current_state();

- /* Registers: FPSIMD-only case */ - BUILD_BUG_ON(SVE_PT_FPSIMD_OFFSET != sizeof(header)); - if ((header.flags & SVE_PT_REGS_MASK) == SVE_PT_REGS_FPSIMD) + BUILD_BUG_ON(SVE_PT_SVE_OFFSET != sizeof(header)); + + switch ((header.flags & SVE_PT_REGS_MASK)) { + case SVE_PT_REGS_FPSIMD: return __fpr_get(target, regset, to);

- /* Otherwise: full SVE case */ + case SVE_PT_REGS_SVE: + start = SVE_PT_SVE_OFFSET; + end = SVE_PT_SVE_FFR_OFFSET(vq) + SVE_PT_SVE_FFR_SIZE(vq); + membuf_write(&to, target->thread.sve_state, end - start);

- BUILD_BUG_ON(SVE_PT_SVE_OFFSET != sizeof(header)); - start = SVE_PT_SVE_OFFSET; - end = SVE_PT_SVE_FFR_OFFSET(vq) + SVE_PT_SVE_FFR_SIZE(vq); - membuf_write(&to, target->thread.sve_state, end - start); + start = end; + end = SVE_PT_SVE_FPSR_OFFSET(vq); + membuf_zero(&to, end - start);

- start = end; - end = SVE_PT_SVE_FPSR_OFFSET(vq); - membuf_zero(&to, end - start); + /* + * Copy fpsr, and fpcr which must follow contiguously in + * struct fpsimd_state: + */ + start = end; + end = SVE_PT_SVE_FPCR_OFFSET(vq) + SVE_PT_SVE_FPCR_SIZE; + membuf_write(&to, &target->thread.uw.fpsimd_state.fpsr, + end - start);

- /* - * Copy fpsr, and fpcr which must follow contiguously in - * struct fpsimd_state: - */ - start = end; - end = SVE_PT_SVE_FPCR_OFFSET(vq) + SVE_PT_SVE_FPCR_SIZE; - membuf_write(&to, &target->thread.uw.fpsimd_state.fpsr, end - start); + start = end; + end = sve_size_from_header(&header); + return membuf_zero(&to, end - start);

- start = end; - end = sve_size_from_header(&header); - return membuf_zero(&to, end - start); + default: + return 0; + } }

-static int sve_set(struct task_struct *target, +static int sve_get(struct task_struct *target, const struct user_regset *regset, - unsigned int pos, unsigned int count, - const void *kbuf, const void __user *ubuf) + struct membuf to) +{ + if (!system_supports_sve()) + return -EINVAL; + + return sve_get_common(target, regset, to, ARM64_VEC_SVE); +} + +static int sve_set_common(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf, + enum vec_type type) { int ret; struct user_sve_header header; unsigned int vq; unsigned long start, end;

- if (!system_supports_sve()) - return -EINVAL; - /* Header */ if (count < sizeof(header)) return -EINVAL; @@ -814,13 +854,37 @@ static int sve_set(struct task_struct *target, * Apart from SVE_PT_REGS_MASK, all SVE_PT_* flags are consumed by * vec_set_vector_length(), which will also validate them for us: */ - ret = vec_set_vector_length(target, ARM64_VEC_SVE, header.vl, + ret = vec_set_vector_length(target, type, header.vl, ((unsigned long)header.flags & ~SVE_PT_REGS_MASK) << 16); if (ret) goto out;

/* Actual VL set may be less than the user asked for: */ - vq = sve_vq_from_vl(task_get_sve_vl(target)); + vq = sve_vq_from_vl(task_get_vl(target, type)); + + /* Enter/exit streaming mode */ + if (system_supports_sme()) { + u64 old_svcr = target->thread.svcr; + + switch (type) { + case ARM64_VEC_SVE: + target->thread.svcr &= ~SYS_SVCR_EL0_SM_MASK; + break; + case ARM64_VEC_SME: + target->thread.svcr |= SYS_SVCR_EL0_SM_MASK; + break; + default: + WARN_ON_ONCE(1); + return -EINVAL; + } + + /* + * If we switched then invalidate any existing SVE + * state and ensure there's storage. + */ + if (target->thread.svcr != old_svcr) + sve_alloc(target); + }

/* Registers: FPSIMD-only case */

@@ -832,7 +896,10 @@ static int sve_set(struct task_struct *target, goto out; }

- /* Otherwise: full SVE case */ + /* + * Otherwise: no registers or full SVE case. For backwards + * compatibility reasons we treat empty flags as SVE registers. + */

/* * If setting a different VL from the requested VL and there is @@ -853,8 +920,9 @@ static int sve_set(struct task_struct *target,

/* * Ensure target->thread.sve_state is up to date with target's - * FPSIMD regs, so that a short copyin leaves trailing registers - * unmodified. + * FPSIMD regs, so that a short copyin leaves trailing + * registers unmodified. Always enable SVE even if going into + * streaming mode. */ fpsimd_sync_to_sve(target); set_tsk_thread_flag(target, TIF_SVE); @@ -890,8 +958,46 @@ static int sve_set(struct task_struct *target, return ret; }

+static int sve_set(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf) +{ + if (!system_supports_sve()) + return -EINVAL; + + return sve_set_common(target, regset, pos, count, kbuf, ubuf, + ARM64_VEC_SVE); +} + #endif /* CONFIG_ARM64_SVE */

+#ifdef CONFIG_ARM64_SME + +static int ssve_get(struct task_struct *target, + const struct user_regset *regset, + struct membuf to) +{ + if (!system_supports_sme()) + return -EINVAL; + + return sve_get_common(target, regset, to, ARM64_VEC_SME); +} + +static int ssve_set(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf) +{ + if (!system_supports_sme()) + return -EINVAL; + + return sve_set_common(target, regset, pos, count, kbuf, ubuf, + ARM64_VEC_SME); +} + +#endif /* CONFIG_ARM64_SME */ + #ifdef CONFIG_ARM64_PTR_AUTH static int pac_mask_get(struct task_struct *target, const struct user_regset *regset, @@ -1109,6 +1215,9 @@ enum aarch64_regset { #ifdef CONFIG_ARM64_SVE REGSET_SVE, #endif +#ifdef CONFIG_ARM64_SVE + REGSET_SSVE, +#endif #ifdef CONFIG_ARM64_PTR_AUTH REGSET_PAC_MASK, REGSET_PAC_ENABLED_KEYS, @@ -1189,6 +1298,17 @@ static const struct user_regset aarch64_regsets[] = { .set = sve_set, }, #endif +#ifdef CONFIG_ARM64_SME + [REGSET_SSVE] = { /* Streaming mode SVE */ + .core_note_type = NT_ARM_SSVE, + .n = DIV_ROUND_UP(SVE_PT_SIZE(SVE_VQ_MAX, SVE_PT_REGS_SVE), + SVE_VQ_BYTES), + .size = SVE_VQ_BYTES, + .align = SVE_VQ_BYTES, + .regset_get = ssve_get, + .set = ssve_set, + }, +#endif #ifdef CONFIG_ARM64_PTR_AUTH [REGSET_PAC_MASK] = { .core_note_type = NT_ARM_PAC_MASK, diff --git a/include/uapi/linux/elf.h b/include/uapi/linux/elf.h index 61bf4774b8f2..61502388683f 100644 --- a/include/uapi/linux/elf.h +++ b/include/uapi/linux/elf.h @@ -427,6 +427,7 @@ typedef struct elf64_shdr { #define NT_ARM_PACG_KEYS 0x408 /* ARM pointer authentication generic key */ #define NT_ARM_TAGGED_ADDR_CTRL 0x409 /* arm64 tagged address control (prctl()) */ #define NT_ARM_PAC_ENABLED_KEYS 0x40a /* arm64 ptr auth enabled keys (prctl()) */ +#define NT_ARM_SSVE 0x40b /* ARM Streaming SVE registers */ #define NT_ARC_V2 0x600 /* ARCv2 accumulator/extra registers */ #define NT_VMCOREDD 0x700 /* Vmcore Device Dump Note */ #define NT_MIPS_DSP 0x800 /* MIPS DSP ASE registers */

The ZA array can be read and written with the NT_ARM_ZA. Similarly to our interface for the SVE vector registers the regset consists of a header with information on the current vector length followed by an optional register data payload, represented as for signals as a series of horizontal vectors from 0 to VL/8 in the endianness independent format used for vectors.

On get if ZA is enabled then register data will be provided, otherwise it will be omitted. On set if register data is provided then ZA is enabled and initialized using the provided data, otherwise it is disabled.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/uapi/asm/ptrace.h | 56 +++++++++++ arch/arm64/kernel/ptrace.c | 144 +++++++++++++++++++++++++++ include/uapi/linux/elf.h | 1 + 3 files changed, 201 insertions(+)

diff --git a/arch/arm64/include/uapi/asm/ptrace.h b/arch/arm64/include/uapi/asm/ptrace.h index 522b925a78c1..7fa2f7036aa7 100644 --- a/arch/arm64/include/uapi/asm/ptrace.h +++ b/arch/arm64/include/uapi/asm/ptrace.h @@ -268,6 +268,62 @@ struct user_pac_generic_keys { __uint128_t apgakey; };

+/* ZA state (NT_ARM_ZA) */ + +struct user_za_header { + __u32 size; /* total meaningful regset content in bytes */ + __u32 max_size; /* maxmium possible size for this thread */ + __u16 vl; /* current vector length */ + __u16 max_vl; /* maximum possible vector length */ + __u16 flags; + __u16 __reserved; +}; + +/* + * Common ZA_PT_* flags: + * These must be kept in sync with prctl interface in <linux/prctl.h> + */ +#define ZA_PT_VL_INHERIT ((1 << 17) /* PR_SME_VL_INHERIT */ >> 16) +#define ZA_PT_VL_ONEXEC ((1 << 18) /* PR_SME_SET_VL_ONEXEC */ >> 16) + + +/* + * The remainder of the ZA state follows struct user_za_header. The + * total size of the ZA state (including header) depends on the + * metadata in the header: ZA_PT_SIZE(vq, flags) gives the total size + * of the state in bytes, including the header. + * + * Refer to <asm/sigcontext.h> for details of how to pass the correct + * "vq" argument to these macros. + */ + +/* Offset from the start of struct user_za_header to the register data */ +#define ZA_PT_ZA_OFFSET \ + ((sizeof(struct user_za_header) + (__SVE_VQ_BYTES - 1)) \ + / __SVE_VQ_BYTES * __SVE_VQ_BYTES) + +/* + * The payload starts at offset ZA_PT_ZA_OFFSET, and is of size + * ZA_PT_ZA_SIZE(vq, flags). + * + * The ZA array is stored as a sequence of horizontal vectors ZAV of SVL/8 + * bytes each, starting from vector 0. + * + * Additional data might be appended in the future. + * + * The ZA matrix is represented in memory in an endianness-invariant layout + * which differs from the layout used for the FPSIMD V-registers on big-endian + * systems: see sigcontext.h for more explanation. + */ + +#define ZA_PT_ZAV_OFFSET(vq, n) \ + (ZA_PT_ZA_OFFSET + ((vq * __SVE_VQ_BYTES) * n)) + +#define ZA_PT_ZA_SIZE(vq) ((vq * __SVE_VQ_BYTES) * (vq * __SVE_VQ_BYTES)) + +#define ZA_PT_SIZE(vq) \ + (ZA_PT_ZA_OFFSET + ZA_PT_ZA_SIZE(vq)) + #endif /* __ASSEMBLY__ */

#endif /* _UAPI__ASM_PTRACE_H */ diff --git a/arch/arm64/kernel/ptrace.c b/arch/arm64/kernel/ptrace.c index 0b8324fa1abc..d6f27d3bf6e7 100644 --- a/arch/arm64/kernel/ptrace.c +++ b/arch/arm64/kernel/ptrace.c @@ -996,6 +996,141 @@ static int ssve_set(struct task_struct *target, ARM64_VEC_SME); }

+static int za_get(struct task_struct *target, + const struct user_regset *regset, + struct membuf to) +{ + struct user_za_header header; + unsigned int vq; + unsigned long start, end; + + if (!system_supports_sme()) + return -EINVAL; + + /* Header */ + memset(&header, 0, sizeof(header)); + + if (test_tsk_thread_flag(target, TIF_SME_VL_INHERIT)) + header.flags |= ZA_PT_VL_INHERIT; + + header.vl = task_get_sme_vl(target); + vq = sve_vq_from_vl(header.vl); + header.max_vl = sme_max_vl(); + header.max_size = ZA_PT_SIZE(vq); + + /* If ZA is not active there is only the header */ + if (thread_za_enabled(&target->thread)) + header.size = ZA_PT_SIZE(vq); + else + header.size = ZA_PT_ZA_OFFSET; + + membuf_write(&to, &header, sizeof(header)); + + BUILD_BUG_ON(ZA_PT_ZA_OFFSET != sizeof(header)); + end = ZA_PT_ZA_OFFSET; +; + if (target == current) + fpsimd_preserve_current_state(); + + /* Any register data to include? */ + if (thread_za_enabled(&target->thread)) { + start = end; + end = ZA_PT_SIZE(vq); + membuf_write(&to, target->thread.za_state, end - start); + } + + /* Zero any trailing padding */ + start = end; + end = ALIGN(header.size, SVE_VQ_BYTES); + return membuf_zero(&to, end - start); +} + +static int za_set(struct task_struct *target, + const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf) +{ + int ret; + struct user_za_header header; + unsigned int vq; + unsigned long start, end; + + if (!system_supports_sme()) + return -EINVAL; + + /* Header */ + if (count < sizeof(header)) + return -EINVAL; + ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &header, + 0, sizeof(header)); + if (ret) + goto out; + + /* + * All current ZA_PT_* flags are consumed by + * vec_set_vector_length(), which will also validate them for + * us: + */ + ret = vec_set_vector_length(target, ARM64_VEC_SME, header.vl, + ((unsigned long)header.flags) << 16); + if (ret) + goto out; + + /* Actual VL set may be less than the user asked for: */ + vq = sve_vq_from_vl(task_get_sme_vl(target)); + + /* Ensure there is some SVE storage for streaming mode */ + if (!target->thread.sve_state) { + sve_alloc(target); + if (!target->thread.sve_state) { + clear_thread_flag(TIF_SME); + ret = -ENOMEM; + goto out; + } + } + + /* Allocate/reinit ZA storage */ + sme_alloc(target); + if (!target->thread.za_state) { + ret = -ENOMEM; + clear_tsk_thread_flag(target, TIF_SME); + goto out; + } + + /* If there is no data then disable ZA */ + if (!count) { + target->thread.svcr &= ~SYS_SVCR_EL0_ZA_MASK; + goto out; + } + + /* + * If setting a different VL from the requested VL and there is + * register data, the data layout will be wrong: don't even + * try to set the registers in this case. + */ + if (vq != sve_vq_from_vl(header.vl)) { + ret = -EIO; + goto out; + } + + BUILD_BUG_ON(ZA_PT_ZA_OFFSET != sizeof(header)); + start = ZA_PT_ZA_OFFSET; + end = ZA_PT_SIZE(vq); + ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, + target->thread.za_state, + start, end); + if (ret) + goto out; + + /* Mark ZA as active and let userspace use it */ + set_tsk_thread_flag(target, TIF_SME); + target->thread.svcr |= SYS_SVCR_EL0_ZA_MASK; + +out: + fpsimd_flush_task_state(target); + return ret; +} + #endif /* CONFIG_ARM64_SME */

#ifdef CONFIG_ARM64_PTR_AUTH @@ -1217,6 +1352,7 @@ enum aarch64_regset { #endif #ifdef CONFIG_ARM64_SVE REGSET_SSVE, + REGSET_ZA, #endif #ifdef CONFIG_ARM64_PTR_AUTH REGSET_PAC_MASK, @@ -1308,6 +1444,14 @@ static const struct user_regset aarch64_regsets[] = { .regset_get = ssve_get, .set = ssve_set, }, + [REGSET_ZA] = { /* SME ZA */ + .core_note_type = NT_ARM_ZA, + .n = DIV_ROUND_UP(ZA_PT_ZA_SIZE(SVE_VQ_MAX), SVE_VQ_BYTES), + .size = SVE_VQ_BYTES, + .align = SVE_VQ_BYTES, + .regset_get = za_get, + .set = za_set, + }, #endif #ifdef CONFIG_ARM64_PTR_AUTH [REGSET_PAC_MASK] = { diff --git a/include/uapi/linux/elf.h b/include/uapi/linux/elf.h index 61502388683f..7ef574f3256a 100644 --- a/include/uapi/linux/elf.h +++ b/include/uapi/linux/elf.h @@ -428,6 +428,7 @@ typedef struct elf64_shdr { #define NT_ARM_TAGGED_ADDR_CTRL 0x409 /* arm64 tagged address control (prctl()) */ #define NT_ARM_PAC_ENABLED_KEYS 0x40a /* arm64 ptr auth enabled keys (prctl()) */ #define NT_ARM_SSVE 0x40b /* ARM Streaming SVE registers */ +#define NT_ARM_ZA 0x40c /* ARM SME ZA registers */ #define NT_ARC_V2 0x600 /* ARCv2 accumulator/extra registers */ #define NT_VMCOREDD 0x700 /* Vmcore Device Dump Note */ #define NT_MIPS_DSP 0x800 /* MIPS DSP ASE registers */

Both streaming mode and ZA may increase power consumption when they are enabled and streaming mode makes many FPSIMD and SVE instructions undefined which will cause problems for any kernel mode floating point so disable both when we flush the CPU state. This covers both kernel_neon_begin() and idle and after flushing the state a reload is always required anyway.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/kernel/fpsimd.c | 9 +++++++++ 1 file changed, 9 insertions(+)

diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c index 2e931373b361..d21d16bb03a8 100644 --- a/arch/arm64/kernel/fpsimd.c +++ b/arch/arm64/kernel/fpsimd.c @@ -1742,6 +1742,15 @@ static void fpsimd_flush_cpu_state(void) { WARN_ON(!system_supports_fpsimd()); __this_cpu_write(fpsimd_last_state.st, NULL); + + /* + * Leaving streaming mode enabled will cause issues for any kernel + * NEON and leaving streaming mode or ZA enabled may increase power + * consumption. + */ + if (system_supports_sme()) + sme_smstop(); + set_thread_flag(TIF_FOREIGN_FPSTATE); }

For the time being we do not support use of SME by KVM guests, support for this will be enabled in future. In order to prevent any side effects or side channels via the new system registers, including the EL0 read/write register TPIDR2, explicitly undefine all the system registers added by SME and mask out the SME bitfield in SYS_ID_AA64PFR1.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/kvm/sys_regs.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c index 4dc2fba316ff..43516eda9143 100644 --- a/arch/arm64/kvm/sys_regs.c +++ b/arch/arm64/kvm/sys_regs.c @@ -1089,6 +1089,8 @@ static u64 read_id_reg(const struct kvm_vcpu *vcpu, case SYS_ID_AA64PFR1_EL1: if (!kvm_has_mte(vcpu->kvm)) val &= ~ARM64_FEATURE_MASK(ID_AA64PFR1_MTE); + + val &= ~ARM64_FEATURE_MASK(ID_AA64PFR1_SME); break; case SYS_ID_AA64ISAR1_EL1: if (!vcpu_has_ptrauth(vcpu)) @@ -1508,7 +1510,7 @@ static const struct sys_reg_desc sys_reg_descs[] = { ID_UNALLOCATED(4,2), ID_UNALLOCATED(4,3), ID_SANITISED(ID_AA64ZFR0_EL1), - ID_UNALLOCATED(4,5), + ID_HIDDEN(ID_AA64SMFR0_EL1), ID_UNALLOCATED(4,6), ID_UNALLOCATED(4,7),

@@ -1551,6 +1553,8 @@ static const struct sys_reg_desc sys_reg_descs[] = {

{ SYS_DESC(SYS_ZCR_EL1), NULL, reset_val, ZCR_EL1, 0, .visibility = sve_visibility }, { SYS_DESC(SYS_TRFCR_EL1), undef_access }, + { SYS_DESC(SYS_SMPRI_EL1), undef_access }, + { SYS_DESC(SYS_SMCR_EL1), undef_access }, { SYS_DESC(SYS_TTBR0_EL1), access_vm_reg, reset_unknown, TTBR0_EL1 }, { SYS_DESC(SYS_TTBR1_EL1), access_vm_reg, reset_unknown, TTBR1_EL1 }, { SYS_DESC(SYS_TCR_EL1), access_vm_reg, reset_val, TCR_EL1, 0 }, @@ -1633,8 +1637,10 @@ static const struct sys_reg_desc sys_reg_descs[] = {

{ SYS_DESC(SYS_CCSIDR_EL1), access_ccsidr }, { SYS_DESC(SYS_CLIDR_EL1), access_clidr }, + { SYS_DESC(SYS_SMIDR_EL1), undef_access }, { SYS_DESC(SYS_CSSELR_EL1), access_csselr, reset_unknown, CSSELR_EL1 }, { SYS_DESC(SYS_CTR_EL0), access_ctr }, + { SYS_DESC(SYS_SVCR_EL0), undef_access },

{ PMU_SYS_REG(SYS_PMCR_EL0), .access = access_pmcr, .reset = reset_pmcr, .reg = PMCR_EL0 }, @@ -1674,6 +1680,7 @@ static const struct sys_reg_desc sys_reg_descs[] = {

{ SYS_DESC(SYS_TPIDR_EL0), NULL, reset_unknown, TPIDR_EL0 }, { SYS_DESC(SYS_TPIDRRO_EL0), NULL, reset_unknown, TPIDRRO_EL0 }, + { SYS_DESC(SYS_TPIDR2_EL0), undef_access },

{ SYS_DESC(SYS_SCXTNUM_EL0), undef_access },

While we don't currently support SME in guests we do currently support it for the host system so we need to take care of SME's impact, including the floating point register state, when running guests. Simiarly to SVE we need to manage the traps in CPACR_RL1, what is new is the handling of streaming mode and ZA.

Normally we defer any handling of the floating point register state until the guest first uses it however if the system is in streaming mode FPSIMD and SVE operations may generate SME traps which we would need to distinguish from actual attempts by the guest to use SME. Rather than do this for the time being if we are in streaming mode when entering the guest we force the floating point state to be saved immediately and exit streaming mode, meaning that the guest won't generate SME traps for supported operations.

We could handle ZA in the access trap similarly to the FPSIMD/SVE state without the disruption caused by streaming mode but for simplicity handle it the same way as streaming mode for now.

This will be revisited when we support SME for guests (hopefully before SME hardware becomes available), for now it will only incur additional cost on systems with SME and even there only if streaming mode or ZA are enabled.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Catalin Marinas catalin.marinas@arm.com --- arch/arm64/include/asm/kvm_host.h | 1 + arch/arm64/kvm/fpsimd.c | 36 +++++++++++++++++++++++++++++++ 2 files changed, 37 insertions(+)

diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h index 7dc85d5a6552..404b7358ba96 100644 --- a/arch/arm64/include/asm/kvm_host.h +++ b/arch/arm64/include/asm/kvm_host.h @@ -438,6 +438,7 @@ struct kvm_vcpu_arch { #define KVM_ARM64_DEBUG_STATE_SAVE_SPE (1 << 12) /* Save SPE context if active */ #define KVM_ARM64_DEBUG_STATE_SAVE_TRBE (1 << 13) /* Save TRBE context if active */ #define KVM_ARM64_FP_FOREIGN_FPSTATE (1 << 14) +#define KVM_ARM64_HOST_SME_ENABLED (1 << 15) /* SME enabled for EL0 */

#define KVM_GUESTDBG_VALID_MASK (KVM_GUESTDBG_ENABLE | \ KVM_GUESTDBG_USE_SW_BP | \ diff --git a/arch/arm64/kvm/fpsimd.c b/arch/arm64/kvm/fpsimd.c index 338733ac63f8..1c585553d74f 100644 --- a/arch/arm64/kvm/fpsimd.c +++ b/arch/arm64/kvm/fpsimd.c @@ -82,6 +82,26 @@ void kvm_arch_vcpu_load_fp(struct kvm_vcpu *vcpu)

if (read_sysreg(cpacr_el1) & CPACR_EL1_ZEN_EL0EN) vcpu->arch.flags |= KVM_ARM64_HOST_SVE_ENABLED; + + /* + * We don't currently support SME guests but if we leave + * things in streaming mode then when the guest starts running + * FPSIMD or SVE code it may generate SME traps so as a + * special case if we are in streaming mode we force the host + * state to be saved now and exit streaming mode so that we + * don't have to handle any SME traps for valid guest + * operations. Do this for ZA as well for now for simplicity. + */ + if (system_supports_sme()) { + if (read_sysreg(cpacr_el1) & CPACR_EL1_SMEN_EL0EN) + vcpu->arch.flags |= KVM_ARM64_HOST_SME_ENABLED; + + if (read_sysreg_s(SYS_SVCR_EL0) & + (SYS_SVCR_EL0_SM_MASK | SYS_SVCR_EL0_ZA_MASK)) { + vcpu->arch.flags &= ~KVM_ARM64_FP_HOST; + fpsimd_save_and_flush_cpu_state(); + } + } }

void kvm_arch_vcpu_ctxflush_fp(struct kvm_vcpu *vcpu) @@ -129,6 +149,22 @@ void kvm_arch_vcpu_put_fp(struct kvm_vcpu *vcpu)

local_irq_save(flags);

+ /* + * If we have VHE then the Hyp code will reset CPACR_EL1 to + * CPACR_EL1_DEFAULT and we need to reenable SME. + */ + if (has_vhe() && system_supports_sme()) { + /* Also restore EL0 state seen on entry */ + if (vcpu->arch.flags & KVM_ARM64_HOST_SME_ENABLED) + sysreg_clear_set(CPACR_EL1, 0, + CPACR_EL1_SMEN_EL0EN | + CPACR_EL1_SMEN_EL1EN); + else + sysreg_clear_set(CPACR_EL1, + CPACR_EL1_SMEN_EL0EN, + CPACR_EL1_SMEN_EL1EN); + } + if (vcpu->arch.flags & KVM_ARM64_FP_ENABLED) { if (vcpu_has_sve(vcpu)) { __vcpu_sys_reg(vcpu, ZCR_EL1) = read_sysreg_el1(SYS_ZCR);

As for the kernel so that we don't have ambitious toolchain requirements to build the tests manually encode some of the SVE instructions.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- tools/testing/selftests/arm64/fp/sme-inst.h | 51 +++++++++++++++++++++ 1 file changed, 51 insertions(+) create mode 100644 tools/testing/selftests/arm64/fp/sme-inst.h

diff --git a/tools/testing/selftests/arm64/fp/sme-inst.h b/tools/testing/selftests/arm64/fp/sme-inst.h new file mode 100644 index 000000000000..7191e53ca1c0 --- /dev/null +++ b/tools/testing/selftests/arm64/fp/sme-inst.h @@ -0,0 +1,51 @@ +// SPDX-License-Identifier: GPL-2.0-only +// Copyright (C) 2021-2 ARM Limited. +// Original author: Mark Brown broonie@kernel.org + +#ifndef SME_INST_H +#define SME_INST_H + +/* + * RDSVL X\nx, #\imm + */ +.macro rdsvl nx, imm + .inst 0x4bf5800 \ + | (\imm << 5) \ + | (\nx) +.endm + +.macro smstop + msr S0_3_C4_C6_3, xzr +.endm + +.macro smstart_za + msr S0_3_C4_C5_3, xzr +.endm + +.macro smstart_sm + msr S0_3_C4_C3_3, xzr +.endm + +/* + * LDR (vector to ZA array): + * LDR ZA[\nw, #\offset], [X\nxbase, #\offset, MUL VL] + */ +.macro _ldr_za nw, nxbase, offset=0 + .inst 0xe1000000 \ + | (((\nw) & 3) << 13) \ + | ((\nxbase) << 5) \ + | ((\offset) & 7) +.endm + +/* + * STR (vector from ZA array): + * STR ZA[\nw, #\offset], [X\nxbase, #\offset, MUL VL] + */ +.macro _str_za nw, nxbase, offset=0 + .inst 0xe1200000 \ + | (((\nw) & 3) << 13) \ + | ((\nxbase) << 5) \ + | ((\offset) & 7) +.endm + +#endif

The Scalable Matrix Extension adds a new system register TPIDR2 intended to be used by libc for its own thread specific use, add some kselftests which exercise the ABI for it.

Since this test should with some adjustment work for TPIDR and any other similar registers added in future add tests for it in a separate directory rather than placing it with the other floating point tests, nothing existing looked suitable so I created a new test directory called "abi".

Since this feature is intended to be used by libc the test is built as freestanding code using nolibc so we don't end up with the test program and libc both trying to manage the register simultaneously and distrupting each other. As a result of being written using nolibc rather than using hwcaps to identify if SME is available in the system we check for the default SME vector length configuration in proc, adding hwcap support to nolibc seems like disproportionate effort and didn't feel entirely idiomatic for what nolibc is trying to do.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- tools/testing/selftests/arm64/abi/.gitignore | 1 + tools/testing/selftests/arm64/abi/Makefile | 9 +- tools/testing/selftests/arm64/abi/tpidr2.c | 298 +++++++++++++++++++ 3 files changed, 307 insertions(+), 1 deletion(-) create mode 100644 tools/testing/selftests/arm64/abi/tpidr2.c

diff --git a/tools/testing/selftests/arm64/abi/.gitignore b/tools/testing/selftests/arm64/abi/.gitignore index b79cf5814c23..b9e54417250d 100644 --- a/tools/testing/selftests/arm64/abi/.gitignore +++ b/tools/testing/selftests/arm64/abi/.gitignore @@ -1 +1,2 @@ syscall-abi +tpidr2 diff --git a/tools/testing/selftests/arm64/abi/Makefile b/tools/testing/selftests/arm64/abi/Makefile index 96eba974ac8d..c8d7f2495eb2 100644 --- a/tools/testing/selftests/arm64/abi/Makefile +++ b/tools/testing/selftests/arm64/abi/Makefile @@ -1,8 +1,15 @@ # SPDX-License-Identifier: GPL-2.0 # Copyright (C) 2021 ARM Limited

-TEST_GEN_PROGS := syscall-abi +TEST_GEN_PROGS := syscall-abi tpidr2

include ../../lib.mk

$(OUTPUT)/syscall-abi: syscall-abi.c syscall-abi-asm.S + +# Build with nolibc since TPIDR2 is intended to be actively managed by +# libc and we're trying to test the functionality that it depends on here. +$(OUTPUT)/tpidr2: tpidr2.c + $(CC) -fno-asynchronous-unwind-tables -fno-ident -s -Os -nostdlib \ + -static -include ../../../../include/nolibc/nolibc.h \ + -ffreestanding -Wall $^ -o $@ -lgcc diff --git a/tools/testing/selftests/arm64/abi/tpidr2.c b/tools/testing/selftests/arm64/abi/tpidr2.c new file mode 100644 index 000000000000..351a098b503a --- /dev/null +++ b/tools/testing/selftests/arm64/abi/tpidr2.c @@ -0,0 +1,298 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#include <linux/sched.h> +#include <linux/wait.h> + +#define SYS_TPIDR2 "S3_3_C13_C0_5" + +#define EXPECTED_TESTS 5 + +static void putstr(const char *str) +{ + write(1, str, strlen(str)); +} + +static void putnum(unsigned int num) +{ + char c; + + if (num / 10) + putnum(num / 10); + + c = '0' + (num % 10); + write(1, &c, 1); +} + +static int tests_run; +static int tests_passed; +static int tests_failed; +static int tests_skipped; + +static void set_tpidr2(uint64_t val) +{ + asm volatile ( + "msr " SYS_TPIDR2 ", %0\n" + : + : "r"(val) + : "cc"); +} + +static uint64_t get_tpidr2(void) +{ + uint64_t val; + + asm volatile ( + "mrs %0, " SYS_TPIDR2 "\n" + : "=r"(val) + : + : "cc"); + + return val; +} + +static void print_summary(void) +{ + if (tests_passed + tests_failed + tests_skipped != EXPECTED_TESTS) + putstr("# UNEXPECTED TEST COUNT: "); + + putstr("# Totals: pass:"); + putnum(tests_passed); + putstr(" fail:"); + putnum(tests_failed); + putstr(" xfail:0 xpass:0 skip:"); + putnum(tests_skipped); + putstr(" error:0\n"); +} + +/* Processes should start with TPIDR2 == 0 */ +static int default_value(void) +{ + return get_tpidr2() == 0; +} + +/* If we set TPIDR2 we should read that value */ +static int write_read(void) +{ + set_tpidr2(getpid()); + + return getpid() == get_tpidr2(); +} + +/* If we set a value we should read the same value after scheduling out */ +static int write_sleep_read(void) +{ + set_tpidr2(getpid()); + + msleep(100); + + return getpid() == get_tpidr2(); +} + +/* + * If we fork the value in the parent should be unchanged and the + * child should start with the same value and be able to set its own + * value. + */ +static int write_fork_read(void) +{ + pid_t newpid, waiting, oldpid; + int status; + + set_tpidr2(getpid()); + + oldpid = getpid(); + newpid = fork(); + if (newpid == 0) { + /* In child */ + if (get_tpidr2() != oldpid) { + putstr("# TPIDR2 changed in child: "); + putnum(get_tpidr2()); + putstr("\n"); + exit(0); + } + + set_tpidr2(getpid()); + if (get_tpidr2() == getpid()) { + exit(1); + } else { + putstr("# Failed to set TPIDR2 in child\n"); + exit(0); + } + } + if (newpid < 0) { + putstr("# fork() failed: -"); + putnum(-newpid); + putstr("\n"); + return 0; + } + + for (;;) { + waiting = waitpid(newpid, &status, 0); + + if (waiting < 0) { + if (errno == EINTR) + continue; + putstr("# waitpid() failed: "); + putnum(errno); + putstr("\n"); + return 0; + } + if (waiting != newpid) { + putstr("# waitpid() returned wrong PID\n"); + return 0; + } + + if (!WIFEXITED(status)) { + putstr("# child did not exit\n"); + return 0; + } + + if (getpid() != get_tpidr2()) { + putstr("# TPIDR2 corrupted in parent\n"); + return 0; + } + + return WEXITSTATUS(status); + } +} + +/* + * sys_clone() has a lot of per architecture variation so just define + * it here rather than adding it to nolibc, plus the raw API is a + * little more convenient for this test. + */ +static int sys_clone(unsigned long clone_flags, unsigned long newsp, + int *parent_tidptr, unsigned long tls, + int *child_tidptr) +{ + return my_syscall5(__NR_clone, clone_flags, newsp, parent_tidptr, tls, + child_tidptr); +} + +/* + * If we clone with CLONE_SETTLS then the value in the parent should + * be unchanged and the child should start with zero and be able to + * set its own value. + */ +static int write_clone_read(void) +{ + int parent_tid, child_tid; + pid_t parent, waiting; + int ret, status; + + parent = getpid(); + set_tpidr2(parent); + + ret = sys_clone(CLONE_SETTLS, 0, &parent_tid, 0, &child_tid); + if (ret == -1) { + putstr("# clone() failed\n"); + putnum(errno); + putstr("\n"); + return 0; + } + + if (ret == 0) { + /* In child */ + if (get_tpidr2() != 0) { + putstr("# TPIDR2 non-zero in child: "); + putnum(get_tpidr2()); + putstr("\n"); + exit(0); + } + + if (gettid() == 0) + putstr("# Child TID==0\n"); + set_tpidr2(gettid()); + if (get_tpidr2() == gettid()) { + exit(1); + } else { + putstr("# Failed to set TPIDR2 in child\n"); + exit(0); + } + } + + for (;;) { + waiting = wait4(ret, &status, __WCLONE, NULL); + + if (waiting < 0) { + if (errno == EINTR) + continue; + putstr("# wait4() failed: "); + putnum(errno); + putstr("\n"); + return 0; + } + if (waiting != ret) { + putstr("# wait4() returned wrong PID "); + putnum(waiting); + putstr("\n"); + return 0; + } + + if (!WIFEXITED(status)) { + putstr("# child did not exit\n"); + return 0; + } + + if (parent != get_tpidr2()) { + putstr("# TPIDR2 corrupted in parent\n"); + return 0; + } + + return WEXITSTATUS(status); + } +} + +#define run_test(name) \ + if (name()) { \ + tests_passed++; \ + } else { \ + tests_failed++; \ + putstr("not "); \ + } \ + putstr("ok "); \ + putnum(++tests_run); \ + putstr(" " #name "\n"); + +int main(int argc, char **argv) +{ + int ret, i; + + putstr("TAP version 13\n"); + putstr("1.."); + putnum(EXPECTED_TESTS); + putstr("\n"); + + putstr("# PID: "); + putnum(getpid()); + putstr("\n"); + + /* + * This test is run with nolibc which doesn't support hwcap and + * it's probably disproportionate to implement so instead check + * for the default vector length configuration in /proc. + */ + ret = open("/proc/sys/abi/sme_default_vector_length", O_RDONLY, 0); + if (ret >= 0) { + run_test(default_value); + run_test(write_read); + run_test(write_sleep_read); + run_test(write_fork_read); + run_test(write_clone_read); + + } else { + putstr("# SME support not present\n"); + + for (i = 0; i < EXPECTED_TESTS; i++) { + putstr("ok "); + putnum(i); + putstr(" skipped, TPIDR2 not supported\n"); + } + + tests_skipped += EXPECTED_TESTS; + } + + print_summary(); + + return 0; +}

One of the features of SME is the addition of streaming mode, in which we have access to a set of streaming mode SVE registers at the SME vector length. Since these are accessed using the SVE instructions let's reuse the existing SVE stress test for testing with a compile time option for controlling the few small differences needed:

- Enter streaming mode immediately on starting the program. - In streaming mode FFR is removed so skip reading and writing FFR.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- tools/testing/selftests/arm64/fp/.gitignore | 1 + tools/testing/selftests/arm64/fp/Makefile | 3 + tools/testing/selftests/arm64/fp/ssve-stress | 59 ++++++++++++++++++++ tools/testing/selftests/arm64/fp/sve-test.S | 20 +++++++ 4 files changed, 83 insertions(+) create mode 100644 tools/testing/selftests/arm64/fp/ssve-stress

diff --git a/tools/testing/selftests/arm64/fp/.gitignore b/tools/testing/selftests/arm64/fp/.gitignore index 6e9a610c5e5d..5729a5b1adfc 100644 --- a/tools/testing/selftests/arm64/fp/.gitignore +++ b/tools/testing/selftests/arm64/fp/.gitignore @@ -5,5 +5,6 @@ rdvl-sve sve-probe-vls sve-ptrace sve-test +ssve-test vec-syscfg vlset diff --git a/tools/testing/selftests/arm64/fp/Makefile b/tools/testing/selftests/arm64/fp/Makefile index a224fff8082b..e6643c9b0474 100644 --- a/tools/testing/selftests/arm64/fp/Makefile +++ b/tools/testing/selftests/arm64/fp/Makefile @@ -5,6 +5,7 @@ TEST_GEN_PROGS := sve-ptrace sve-probe-vls vec-syscfg TEST_PROGS_EXTENDED := fp-pidbench fpsimd-test fpsimd-stress \ rdvl-sme rdvl-sve \ sve-test sve-stress \ + ssve-test ssve-stress \ vlset

all: $(TEST_GEN_PROGS) $(TEST_PROGS_EXTENDED) @@ -19,6 +20,8 @@ sve-ptrace: sve-ptrace.o sve-probe-vls: sve-probe-vls.o rdvl.o sve-test: sve-test.o asm-utils.o $(CC) -nostdlib $^ -o $@ +ssve-test: sve-test.S asm-utils.o + $(CC) -DSSVE -nostdlib $^ -o $@ vec-syscfg: vec-syscfg.o rdvl.o vlset: vlset.o

diff --git a/tools/testing/selftests/arm64/fp/ssve-stress b/tools/testing/selftests/arm64/fp/ssve-stress new file mode 100644 index 000000000000..e2bd2cc184ad --- /dev/null +++ b/tools/testing/selftests/arm64/fp/ssve-stress @@ -0,0 +1,59 @@ +#!/bin/bash +# SPDX-License-Identifier: GPL-2.0-only +# Copyright (C) 2015-2019 ARM Limited. +# Original author: Dave Martin Dave.Martin@arm.com + +set -ue + +NR_CPUS=`nproc` + +pids= +logs= + +cleanup () { + trap - INT TERM CHLD + set +e + + if [ -n "$pids" ]; then + kill $pids + wait $pids + pids= + fi + + if [ -n "$logs" ]; then + cat $logs + rm $logs + logs= + fi +} + +interrupt () { + cleanup + exit 0 +} + +child_died () { + cleanup + exit 1 +} + +trap interrupt INT TERM EXIT + +for x in `seq 0 $((NR_CPUS * 4))`; do + log=`mktemp` + logs=$logs\ $log + ./ssve-test >$log & + pids=$pids\ $! +done + +# Wait for all child processes to be created: +sleep 10 + +while :; do + kill -USR1 $pids +done & +pids=$pids\ $! + +wait + +exit 1 diff --git a/tools/testing/selftests/arm64/fp/sve-test.S b/tools/testing/selftests/arm64/fp/sve-test.S index f5b1b48ffff2..589264231a2d 100644 --- a/tools/testing/selftests/arm64/fp/sve-test.S +++ b/tools/testing/selftests/arm64/fp/sve-test.S @@ -13,6 +13,7 @@ #include <asm/unistd.h> #include "assembler.h" #include "asm-offsets.h" +#include "sme-inst.h"

#define NZR 32 #define NPR 16 @@ -156,6 +157,7 @@ endfunction // We fill the upper lanes of FFR with zeros. // Beware: corrupts P0. function setup_ffr +#ifndef SSVE mov x4, x30

and w0, w0, #0x3 @@ -178,6 +180,9 @@ function setup_ffr wrffr p0.b

ret x4 +#else + ret +#endif endfunction

// Trivial memory compare: compare x2 bytes starting at address x0 with @@ -260,6 +265,7 @@ endfunction // Beware -- corrupts P0. // Clobbers x0-x5. function check_ffr +#ifndef SSVE mov x3, x30

ldr x4, =scratch @@ -280,6 +286,9 @@ function check_ffr mov x2, x5 mov x30, x3 b memcmp +#else + ret +#endif endfunction

// Any SVE register modified here can cause corruption in the main @@ -295,10 +304,12 @@ function irritator_handler movi v0.8b, #1 movi v9.16b, #2 movi v31.8b, #3 +#ifndef SSVE // And P0 rdffr p0.b // And FFR wrffr p15.b +#endif

ret endfunction @@ -359,6 +370,11 @@ endfunction .globl _start function _start _start: +#ifdef SSVE + puts "Streaming mode " + smstart_sm +#endif + // Sanity-check and report the vector length

rdvl x19, #8 @@ -407,6 +423,10 @@ _start: orr w2, w2, #SA_NODEFER bl setsignal

+#ifdef SSVE + smstart_sm // syscalls will have exited streaming mode +#endif + mov x22, #0 // generation number, increments per iteration .Ltest_loop: rdvl x0, #8

Add a stress test for context switching of the ZA register state based on the similar tests Dave Martin wrote for FPSIMD and SVE registers. The test loops indefinitely writing a data pattern to ZA then reading it back and verifying that it's what was expected.

Unlike the other tests we manually assemble the SME instructions since at present no released toolchain has SME support integrated.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- tools/testing/selftests/arm64/fp/.gitignore | 1 + tools/testing/selftests/arm64/fp/Makefile | 3 + tools/testing/selftests/arm64/fp/za-stress | 59 +++ tools/testing/selftests/arm64/fp/za-test.S | 388 ++++++++++++++++++++ 4 files changed, 451 insertions(+) create mode 100644 tools/testing/selftests/arm64/fp/za-stress create mode 100644 tools/testing/selftests/arm64/fp/za-test.S

diff --git a/tools/testing/selftests/arm64/fp/.gitignore b/tools/testing/selftests/arm64/fp/.gitignore index 5729a5b1adfc..ead3197e720b 100644 --- a/tools/testing/selftests/arm64/fp/.gitignore +++ b/tools/testing/selftests/arm64/fp/.gitignore @@ -8,3 +8,4 @@ sve-test ssve-test vec-syscfg vlset +za-test diff --git a/tools/testing/selftests/arm64/fp/Makefile b/tools/testing/selftests/arm64/fp/Makefile index e6643c9b0474..38d2d0d5a0eb 100644 --- a/tools/testing/selftests/arm64/fp/Makefile +++ b/tools/testing/selftests/arm64/fp/Makefile @@ -6,6 +6,7 @@ TEST_PROGS_EXTENDED := fp-pidbench fpsimd-test fpsimd-stress \ rdvl-sme rdvl-sve \ sve-test sve-stress \ ssve-test ssve-stress \ + za-test za-stress \ vlset

all: $(TEST_GEN_PROGS) $(TEST_PROGS_EXTENDED) @@ -24,5 +25,7 @@ ssve-test: sve-test.S asm-utils.o $(CC) -DSSVE -nostdlib $^ -o $@ vec-syscfg: vec-syscfg.o rdvl.o vlset: vlset.o +za-test: za-test.o asm-utils.o + $(CC) -nostdlib $^ -o $@

include ../../lib.mk diff --git a/tools/testing/selftests/arm64/fp/za-stress b/tools/testing/selftests/arm64/fp/za-stress new file mode 100644 index 000000000000..5ac386b55b95 --- /dev/null +++ b/tools/testing/selftests/arm64/fp/za-stress @@ -0,0 +1,59 @@ +#!/bin/bash +# SPDX-License-Identifier: GPL-2.0-only +# Copyright (C) 2015-2019 ARM Limited. +# Original author: Dave Martin Dave.Martin@arm.com + +set -ue + +NR_CPUS=`nproc` + +pids= +logs= + +cleanup () { + trap - INT TERM CHLD + set +e + + if [ -n "$pids" ]; then + kill $pids + wait $pids + pids= + fi + + if [ -n "$logs" ]; then + cat $logs + rm $logs + logs= + fi +} + +interrupt () { + cleanup + exit 0 +} + +child_died () { + cleanup + exit 1 +} + +trap interrupt INT TERM EXIT + +for x in `seq 0 $((NR_CPUS * 4))`; do + log=`mktemp` + logs=$logs\ $log + ./za-test >$log & + pids=$pids\ $! +done + +# Wait for all child processes to be created: +sleep 10 + +while :; do + kill -USR1 $pids +done & +pids=$pids\ $! + +wait + +exit 1 diff --git a/tools/testing/selftests/arm64/fp/za-test.S b/tools/testing/selftests/arm64/fp/za-test.S new file mode 100644 index 000000000000..9ab6f9cd9623 --- /dev/null +++ b/tools/testing/selftests/arm64/fp/za-test.S @@ -0,0 +1,388 @@ +// SPDX-License-Identifier: GPL-2.0-only +// Copyright (C) 2021 ARM Limited. +// Original author: Mark Brown broonie@kernel.org +// +// Scalable Matrix Extension ZA context switch test +// Repeatedly writes unique test patterns into each ZA tile +// and reads them back to verify integrity. +// +// for x in `seq 1 NR_CPUS`; do sve-test & pids=$pids\ $! ; done +// (leave it running for as long as you want...) +// kill $pids + +#include <asm/unistd.h> +#include "assembler.h" +#include "asm-offsets.h" +#include "sme-inst.h" + +.arch_extension sve + +#define MAXVL 2048 +#define MAXVL_B (MAXVL / 8) + +// Declare some storage space to shadow ZA register contents and a +// scratch buffer for a vector. +.pushsection .text +.data +.align 4 +zaref: + .space MAXVL_B * MAXVL_B +scratch: + .space MAXVL_B +.popsection + +// Trivial memory copy: copy x2 bytes, starting at address x1, to address x0. +// Clobbers x0-x3 +function memcpy + cmp x2, #0 + b.eq 1f +0: ldrb w3, [x1], #1 + strb w3, [x0], #1 + subs x2, x2, #1 + b.ne 0b +1: ret +endfunction + +// Generate a test pattern for storage in ZA +// x0: pid +// x1: row in ZA +// x2: generation + +// These values are used to constuct a 32-bit pattern that is repeated in the +// scratch buffer as many times as will fit: +// bits 31:28 generation number (increments once per test_loop) +// bits 27:16 pid +// bits 15: 8 row number +// bits 7: 0 32-bit lane index + +function pattern + mov w3, wzr + bfi w3, w0, #16, #12 // PID + bfi w3, w1, #8, #8 // Row + bfi w3, w2, #28, #4 // Generation + + ldr x0, =scratch + mov w1, #MAXVL_B / 4 + +0: str w3, [x0], #4 + add w3, w3, #1 // Lane + subs w1, w1, #1 + b.ne 0b + + ret +endfunction + +// Get the address of shadow data for ZA horizontal vector xn +.macro _adrza xd, xn, nrtmp + ldr \xd, =zaref + rdsvl \nrtmp, 1 + madd \xd, x\nrtmp, \xn, \xd +.endm + +// Set up test pattern in a ZA horizontal vector +// x0: pid +// x1: row number +// x2: generation +function setup_za + mov x4, x30 + mov x12, x1 // Use x12 for vector select + + bl pattern // Get pattern in scratch buffer + _adrza x0, x12, 2 // Shadow buffer pointer to x0 and x5 + mov x5, x0 + ldr x1, =scratch + bl memcpy // length set up in x2 by _adrza + + _ldr_za 12, 5 // load vector w12 from pointer x5 + + ret x4 +endfunction + +// Trivial memory compare: compare x2 bytes starting at address x0 with +// bytes starting at address x1. +// Returns only if all bytes match; otherwise, the program is aborted. +// Clobbers x0-x5. +function memcmp + cbz x2, 2f + + stp x0, x1, [sp, #-0x20]! + str x2, [sp, #0x10] + + mov x5, #0 +0: ldrb w3, [x0, x5] + ldrb w4, [x1, x5] + add x5, x5, #1 + cmp w3, w4 + b.ne 1f + subs x2, x2, #1 + b.ne 0b + +1: ldr x2, [sp, #0x10] + ldp x0, x1, [sp], #0x20 + b.ne barf + +2: ret +endfunction + +// Verify that a ZA vector matches its shadow in memory, else abort +// x0: row number +// Clobbers x0-x7 and x12. +function check_za + mov x3, x30 + + mov x12, x0 + _adrza x5, x0, 6 // pointer to expected value in x5 + mov x4, x0 + ldr x7, =scratch // x7 is scratch + + mov x0, x7 // Poison scratch + mov x1, x6 + bl memfill_ae + + _str_za 12, 7 // save vector w12 to pointer x7 + + mov x0, x5 + mov x1, x7 + mov x2, x6 + mov x30, x3 + b memcmp +endfunction + +// Any SME register modified here can cause corruption in the main +// thread -- but *only* the locations modified here. +function irritator_handler + // Increment the irritation signal count (x23): + ldr x0, [x2, #ucontext_regs + 8 * 23] + add x0, x0, #1 + str x0, [x2, #ucontext_regs + 8 * 23] + + // Corrupt some random ZA data +#if 0 + adr x0, .text + (irritator_handler - .text) / 16 * 16 + movi v0.8b, #1 + movi v9.16b, #2 + movi v31.8b, #3 +#endif + + ret +endfunction + +function terminate_handler + mov w21, w0 + mov x20, x2 + + puts "Terminated by signal " + mov w0, w21 + bl putdec + puts ", no error, iterations=" + ldr x0, [x20, #ucontext_regs + 8 * 22] + bl putdec + puts ", signals=" + ldr x0, [x20, #ucontext_regs + 8 * 23] + bl putdecn + + mov x0, #0 + mov x8, #__NR_exit + svc #0 +endfunction + +// w0: signal number +// x1: sa_action +// w2: sa_flags +// Clobbers x0-x6,x8 +function setsignal + str x30, [sp, #-((sa_sz + 15) / 16 * 16 + 16)]! + + mov w4, w0 + mov x5, x1 + mov w6, w2 + + add x0, sp, #16 + mov x1, #sa_sz + bl memclr + + mov w0, w4 + add x1, sp, #16 + str w6, [x1, #sa_flags] + str x5, [x1, #sa_handler] + mov x2, #0 + mov x3, #sa_mask_sz + mov x8, #__NR_rt_sigaction + svc #0 + + cbz w0, 1f + + puts "sigaction failure\n" + b .Labort + +1: ldr x30, [sp], #((sa_sz + 15) / 16 * 16 + 16) + ret +endfunction + +// Main program entry point +.globl _start +function _start +_start: + puts "Streaming mode " + smstart_za + + // Sanity-check and report the vector length + + rdsvl 19, 8 + cmp x19, #128 + b.lo 1f + cmp x19, #2048 + b.hi 1f + tst x19, #(8 - 1) + b.eq 2f + +1: puts "bad vector length: " + mov x0, x19 + bl putdecn + b .Labort + +2: puts "vector length:\t" + mov x0, x19 + bl putdec + puts " bits\n" + + // Obtain our PID, to ensure test pattern uniqueness between processes + mov x8, #__NR_getpid + svc #0 + mov x20, x0 + + puts "PID:\t" + mov x0, x20 + bl putdecn + + mov x23, #0 // Irritation signal count + + mov w0, #SIGINT + adr x1, terminate_handler + mov w2, #SA_SIGINFO + bl setsignal + + mov w0, #SIGTERM + adr x1, terminate_handler + mov w2, #SA_SIGINFO + bl setsignal + + mov w0, #SIGUSR1 + adr x1, irritator_handler + mov w2, #SA_SIGINFO + orr w2, w2, #SA_NODEFER + bl setsignal + + mov x22, #0 // generation number, increments per iteration +.Ltest_loop: + rdsvl 0, 8 + cmp x0, x19 + b.ne vl_barf + + rdsvl 21, 1 // Set up ZA & shadow with test pattern +0: mov x0, x20 + sub x1, x21, #1 + mov x2, x22 + bl setup_za + subs x21, x21, #1 + b.ne 0b + + and x8, x22, #127 // Every 128 interations... + cbz x8, 0f + mov x8, #__NR_getpid // (otherwise minimal syscall) + b 1f +0: + mov x8, #__NR_sched_yield // ...encourage preemption +1: + svc #0 + + mrs x0, S3_3_C4_C2_2 // SVCR should have ZA=1,SM=0 + and x1, x0, #3 + cmp x1, #2 + b.ne svcr_barf + + rdsvl 21, 1 // Verify that the data made it through + rdsvl 24, 1 // Verify that the data made it through +0: sub x0, x24, x21 + bl check_za + subs x21, x21, #1 + bne 0b + + add x22, x22, #1 // Everything still working + b .Ltest_loop + +.Labort: + mov x0, #0 + mov x1, #SIGABRT + mov x8, #__NR_kill + svc #0 +endfunction + +function barf +// fpsimd.c acitivty log dump hack +// ldr w0, =0xdeadc0de +// mov w8, #__NR_exit +// svc #0 +// end hack + smstop + mov x10, x0 // expected data + mov x11, x1 // actual data + mov x12, x2 // data size + + puts "Mismatch: PID=" + mov x0, x20 + bl putdec + puts ", iteration=" + mov x0, x22 + bl putdec + puts ", row=" + mov x0, x21 + bl putdecn + puts "\tExpected [" + mov x0, x10 + mov x1, x12 + bl dumphex + puts "]\n\tGot [" + mov x0, x11 + mov x1, x12 + bl dumphex + puts "]\n" + + mov x8, #__NR_getpid + svc #0 +// fpsimd.c acitivty log dump hack +// ldr w0, =0xdeadc0de +// mov w8, #__NR_exit +// svc #0 +// ^ end of hack + mov x1, #SIGABRT + mov x8, #__NR_kill + svc #0 +// mov x8, #__NR_exit +// mov x1, #1 +// svc #0 +endfunction + +function vl_barf + mov x10, x0 + + puts "Bad active VL: " + mov x0, x10 + bl putdecn + + mov x8, #__NR_exit + mov x1, #1 + svc #0 +endfunction + +function svcr_barf + mov x10, x0 + + puts "Bad SVCR: " + mov x0, x10 + bl putdecn + + mov x8, #__NR_exit + mov x1, #1 + svc #0 +endfunction

In order to allow ptrace of streaming mode SVE registers we have added a new regset for streaming mode which in isolation offers the same ABI as regular SVE with a different vector type. Add this to the array of regsets we handle, together with additional tests for the interoperation of the two regsets.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- tools/testing/selftests/arm64/fp/sve-ptrace.c | 11 +++++++++++ 1 file changed, 11 insertions(+)

diff --git a/tools/testing/selftests/arm64/fp/sve-ptrace.c b/tools/testing/selftests/arm64/fp/sve-ptrace.c index 4bd333768cc4..509b875cb7bb 100644 --- a/tools/testing/selftests/arm64/fp/sve-ptrace.c +++ b/tools/testing/selftests/arm64/fp/sve-ptrace.c @@ -26,6 +26,10 @@ #define NT_ARM_SVE 0x405 #endif

+#ifndef NT_ARM_SSVE +#define NT_ARM_SSVE 0x40b +#endif + struct vec_type { const char *name; unsigned long hwcap_type; @@ -42,6 +46,13 @@ static const struct vec_type vec_types[] = { .regset = NT_ARM_SVE, .prctl_set = PR_SVE_SET_VL, }, + { + .name = "Streaming SVE", + .hwcap_type = AT_HWCAP2, + .hwcap = HWCAP2_SME, + .regset = NT_ARM_SSVE, + .prctl_set = PR_SME_SET_VL, + }, };

#define VL_TESTS (((SVE_VQ_MAX - SVE_VQ_MIN) + 1) * 3)

For every possible combination of SVE and SME vector length verify that for each possible value of SVCR after a syscall we leave streaming mode and ZA is preserved. We don't need to take account of any streaming/non streaming SVE vector length changes in the assembler code since the store instructions will handle the vector length for us. We log if the system supports FA64 and only try to set FFR in streaming mode if it does.

Signed-off-by: Mark Brown broonie@kernel.org Reviewed-by: Shuah Khan skhan@linuxfoundation.org Acked-by: Catalin Marinas catalin.marinas@arm.com --- .../selftests/arm64/abi/syscall-abi-asm.S | 79 ++++++- .../testing/selftests/arm64/abi/syscall-abi.c | 204 ++++++++++++++++-- .../testing/selftests/arm64/abi/syscall-abi.h | 15 ++ 3 files changed, 275 insertions(+), 23 deletions(-) create mode 100644 tools/testing/selftests/arm64/abi/syscall-abi.h

diff --git a/tools/testing/selftests/arm64/abi/syscall-abi-asm.S b/tools/testing/selftests/arm64/abi/syscall-abi-asm.S index 983467cfcee0..b523c21c2278 100644 --- a/tools/testing/selftests/arm64/abi/syscall-abi-asm.S +++ b/tools/testing/selftests/arm64/abi/syscall-abi-asm.S @@ -9,15 +9,42 @@ // invoked is configured in x8 of the input GPR data. // // x0: SVE VL, 0 for FP only +// x1: SME VL // // GPRs: gpr_in, gpr_out // FPRs: fpr_in, fpr_out // Zn: z_in, z_out // Pn: p_in, p_out // FFR: ffr_in, ffr_out +// ZA: za_in, za_out +// SVCR: svcr_in, svcr_out + +#include "syscall-abi.h"

.arch_extension sve

+/* + * LDR (vector to ZA array): + * LDR ZA[\nw, #\offset], [X\nxbase, #\offset, MUL VL] + */ +.macro _ldr_za nw, nxbase, offset=0 + .inst 0xe1000000 \ + | (((\nw) & 3) << 13) \ + | ((\nxbase) << 5) \ + | ((\offset) & 7) +.endm + +/* + * STR (vector from ZA array): + * STR ZA[\nw, #\offset], [X\nxbase, #\offset, MUL VL] + */ +.macro _str_za nw, nxbase, offset=0 + .inst 0xe1200000 \ + | (((\nw) & 3) << 13) \ + | ((\nxbase) << 5) \ + | ((\offset) & 7) +.endm + .globl do_syscall do_syscall: // Store callee saved registers x19-x29 (80 bytes) plus x0 and x1 @@ -30,6 +57,24 @@ do_syscall: stp x25, x26, [sp, #80] stp x27, x28, [sp, #96]

+ // Set SVCR if we're doing SME + cbz x1, 1f + adrp x2, svcr_in + ldr x2, [x2, :lo12:svcr_in] + msr S3_3_C4_C2_2, x2 +1: + + // Load ZA if it's enabled - uses x12 as scratch due to SME LDR + tbz x2, #SVCR_ZA_SHIFT, 1f + mov w12, #0 + ldr x2, =za_in +2: _ldr_za 12, 2 + add x2, x2, x1 + add x12, x12, #1 + cmp x1, x12 + bne 2b +1: + // Load GPRs x8-x28, and save our SP/FP for later comparison ldr x2, =gpr_in add x2, x2, #64 @@ -68,7 +113,7 @@ do_syscall: ldp q30, q31, [x2, #16 * 30] 1:

- // Load the SVE registers if we're doing SVE + // Load the SVE registers if we're doing SVE/SME cbz x0, 1f

ldr x2, =z_in @@ -105,9 +150,14 @@ do_syscall: ldr z30, [x2, #30, MUL VL] ldr z31, [x2, #31, MUL VL]

+ // Only set a non-zero FFR, test patterns must be zero since the + // syscall should clear it - this lets us handle FA64. ldr x2, =ffr_in ldr p0, [x2, #0] + ldr x2, [x2, #0] + cbz x2, 2f wrffr p0.b +2:

ldr x2, =p_in ldr p0, [x2, #0, MUL VL] @@ -169,6 +219,24 @@ do_syscall: stp q28, q29, [x2, #16 * 28] stp q30, q31, [x2, #16 * 30]

+ // Save SVCR if we're doing SME + cbz x1, 1f + mrs x2, S3_3_C4_C2_2 + adrp x3, svcr_out + str x2, [x3, :lo12:svcr_out] +1: + + // Save ZA if it's enabled - uses x12 as scratch due to SME STR + tbz x2, #SVCR_ZA_SHIFT, 1f + mov w12, #0 + ldr x2, =za_out +2: _str_za 12, 2 + add x2, x2, x1 + add x12, x12, #1 + cmp x1, x12 + bne 2b +1: + // Save the SVE state if we have some cbz x0, 1f

@@ -224,6 +292,10 @@ do_syscall: str p14, [x2, #14, MUL VL] str p15, [x2, #15, MUL VL]

+ // Only save FFR if we wrote a value for SME + ldr x2, =ffr_in + ldr x2, [x2, #0] + cbz x2, 1f ldr x2, =ffr_out rdffr p0.b str p0, [x2, #0] @@ -237,4 +309,9 @@ do_syscall: ldp x27, x28, [sp, #96] ldp x29, x30, [sp], #112

+ // Clear SVCR if we were doing SME so future tests don't have ZA + cbz x1, 1f + msr S3_3_C4_C2_2, xzr +1: + ret diff --git a/tools/testing/selftests/arm64/abi/syscall-abi.c b/tools/testing/selftests/arm64/abi/syscall-abi.c index 1e13b7523918..b632bfe9e022 100644 --- a/tools/testing/selftests/arm64/abi/syscall-abi.c +++ b/tools/testing/selftests/arm64/abi/syscall-abi.c @@ -18,9 +18,13 @@

#include "../../kselftest.h"

+#include "syscall-abi.h" + #define NUM_VL ((SVE_VQ_MAX - SVE_VQ_MIN) + 1)

-extern void do_syscall(int sve_vl); +static int default_sme_vl; + +extern void do_syscall(int sve_vl, int sme_vl);

static void fill_random(void *buf, size_t size) { @@ -48,14 +52,15 @@ static struct syscall_cfg { uint64_t gpr_in[NUM_GPR]; uint64_t gpr_out[NUM_GPR];

-static void setup_gpr(struct syscall_cfg *cfg, int sve_vl) +static void setup_gpr(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { fill_random(gpr_in, sizeof(gpr_in)); gpr_in[8] = cfg->syscall_nr; memset(gpr_out, 0, sizeof(gpr_out)); }

-static int check_gpr(struct syscall_cfg *cfg, int sve_vl) +static int check_gpr(struct syscall_cfg *cfg, int sve_vl, int sme_vl, uint64_t svcr) { int errors = 0; int i; @@ -79,13 +84,15 @@ static int check_gpr(struct syscall_cfg *cfg, int sve_vl) uint64_t fpr_in[NUM_FPR * 2]; uint64_t fpr_out[NUM_FPR * 2];

-static void setup_fpr(struct syscall_cfg *cfg, int sve_vl) +static void setup_fpr(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { fill_random(fpr_in, sizeof(fpr_in)); memset(fpr_out, 0, sizeof(fpr_out)); }

-static int check_fpr(struct syscall_cfg *cfg, int sve_vl) +static int check_fpr(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { int errors = 0; int i; @@ -109,13 +116,15 @@ static uint8_t z_zero[__SVE_ZREG_SIZE(SVE_VQ_MAX)]; uint8_t z_in[SVE_NUM_PREGS * __SVE_ZREG_SIZE(SVE_VQ_MAX)]; uint8_t z_out[SVE_NUM_PREGS * __SVE_ZREG_SIZE(SVE_VQ_MAX)];

-static void setup_z(struct syscall_cfg *cfg, int sve_vl) +static void setup_z(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { fill_random(z_in, sizeof(z_in)); fill_random(z_out, sizeof(z_out)); }

-static int check_z(struct syscall_cfg *cfg, int sve_vl) +static int check_z(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { size_t reg_size = sve_vl; int errors = 0; @@ -126,13 +135,17 @@ static int check_z(struct syscall_cfg *cfg, int sve_vl)

/* * After a syscall the low 128 bits of the Z registers should - * be preserved and the rest be zeroed or preserved. + * be preserved and the rest be zeroed or preserved, except if + * we were in streaming mode in which case the low 128 bits may + * also be cleared by the transition out of streaming mode. */ for (i = 0; i < SVE_NUM_ZREGS; i++) { void *in = &z_in[reg_size * i]; void *out = &z_out[reg_size * i];

- if (memcmp(in, out, SVE_VQ_BYTES) != 0) { + if ((memcmp(in, out, SVE_VQ_BYTES) != 0) && + !((svcr & SVCR_SM_MASK) && + memcmp(z_zero, out, SVE_VQ_BYTES) == 0)) { ksft_print_msg("%s SVE VL %d Z%d low 128 bits changed\n", cfg->name, sve_vl, i); errors++; @@ -145,13 +158,15 @@ static int check_z(struct syscall_cfg *cfg, int sve_vl) uint8_t p_in[SVE_NUM_PREGS * __SVE_PREG_SIZE(SVE_VQ_MAX)]; uint8_t p_out[SVE_NUM_PREGS * __SVE_PREG_SIZE(SVE_VQ_MAX)];

-static void setup_p(struct syscall_cfg *cfg, int sve_vl) +static void setup_p(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { fill_random(p_in, sizeof(p_in)); fill_random(p_out, sizeof(p_out)); }

-static int check_p(struct syscall_cfg *cfg, int sve_vl) +static int check_p(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { size_t reg_size = sve_vq_from_vl(sve_vl) * 2; /* 1 bit per VL byte */

@@ -175,8 +190,19 @@ static int check_p(struct syscall_cfg *cfg, int sve_vl) uint8_t ffr_in[__SVE_PREG_SIZE(SVE_VQ_MAX)]; uint8_t ffr_out[__SVE_PREG_SIZE(SVE_VQ_MAX)];

-static void setup_ffr(struct syscall_cfg *cfg, int sve_vl) +static void setup_ffr(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { + /* + * If we are in streaming mode and do not have FA64 then FFR + * is unavailable. + */ + if ((svcr & SVCR_SM_MASK) && + !(getauxval(AT_HWCAP2) & HWCAP2_SME_FA64)) { + memset(&ffr_in, 0, sizeof(ffr_in)); + return; + } + /* * It is only valid to set a contiguous set of bits starting * at 0. For now since we're expecting this to be cleared by @@ -186,7 +212,8 @@ static void setup_ffr(struct syscall_cfg *cfg, int sve_vl) fill_random(ffr_out, sizeof(ffr_out)); }

-static int check_ffr(struct syscall_cfg *cfg, int sve_vl) +static int check_ffr(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { size_t reg_size = sve_vq_from_vl(sve_vl) * 2; /* 1 bit per VL byte */ int errors = 0; @@ -195,6 +222,10 @@ static int check_ffr(struct syscall_cfg *cfg, int sve_vl) if (!sve_vl) return 0;

+ if ((svcr & SVCR_SM_MASK) && + !(getauxval(AT_HWCAP2) & HWCAP2_SME_FA64)) + return 0; + /* After a syscall the P registers should be preserved or zeroed */ for (i = 0; i < reg_size; i++) if (ffr_out[i] && (ffr_in[i] != ffr_out[i])) @@ -206,8 +237,65 @@ static int check_ffr(struct syscall_cfg *cfg, int sve_vl) return errors; }

-typedef void (*setup_fn)(struct syscall_cfg *cfg, int sve_vl); -typedef int (*check_fn)(struct syscall_cfg *cfg, int sve_vl); +uint64_t svcr_in, svcr_out; + +static void setup_svcr(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) +{ + svcr_in = svcr; +} + +static int check_svcr(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) +{ + int errors = 0; + + if (svcr_out & SVCR_SM_MASK) { + ksft_print_msg("%s Still in SM, SVCR %llx\n", + cfg->name, svcr_out); + errors++; + } + + if ((svcr_in & SVCR_ZA_MASK) != (svcr_out & SVCR_ZA_MASK)) { + ksft_print_msg("%s PSTATE.ZA changed, SVCR %llx != %llx\n", + cfg->name, svcr_in, svcr_out); + errors++; + } + + return errors; +} + +uint8_t za_in[SVE_NUM_PREGS * __SVE_ZREG_SIZE(SVE_VQ_MAX)]; +uint8_t za_out[SVE_NUM_PREGS * __SVE_ZREG_SIZE(SVE_VQ_MAX)]; + +static void setup_za(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) +{ + fill_random(za_in, sizeof(za_in)); + memset(za_out, 0, sizeof(za_out)); +} + +static int check_za(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) +{ + size_t reg_size = sme_vl * sme_vl; + int errors = 0; + + if (!(svcr & SVCR_ZA_MASK)) + return 0; + + if (memcmp(za_in, za_out, reg_size) != 0) { + ksft_print_msg("SME VL %d ZA does not match\n", sme_vl); + errors++; + } + + return errors; +} + +typedef void (*setup_fn)(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr); +typedef int (*check_fn)(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr);

/* * Each set of registers has a setup function which is called before @@ -225,20 +313,23 @@ static struct { { setup_z, check_z }, { setup_p, check_p }, { setup_ffr, check_ffr }, + { setup_svcr, check_svcr }, + { setup_za, check_za }, };

-static bool do_test(struct syscall_cfg *cfg, int sve_vl) +static bool do_test(struct syscall_cfg *cfg, int sve_vl, int sme_vl, + uint64_t svcr) { int errors = 0; int i;

for (i = 0; i < ARRAY_SIZE(regset); i++) - regset[i].setup(cfg, sve_vl); + regset[i].setup(cfg, sve_vl, sme_vl, svcr);

- do_syscall(sve_vl); + do_syscall(sve_vl, sme_vl);

for (i = 0; i < ARRAY_SIZE(regset); i++) - errors += regset[i].check(cfg, sve_vl); + errors += regset[i].check(cfg, sve_vl, sme_vl, svcr);

return errors == 0; } @@ -246,9 +337,10 @@ static bool do_test(struct syscall_cfg *cfg, int sve_vl) static void test_one_syscall(struct syscall_cfg *cfg) { int sve_vq, sve_vl; + int sme_vq, sme_vl;

/* FPSIMD only case */ - ksft_test_result(do_test(cfg, 0), + ksft_test_result(do_test(cfg, 0, default_sme_vl, 0), "%s FPSIMD\n", cfg->name);

if (!(getauxval(AT_HWCAP) & HWCAP_SVE)) @@ -265,8 +357,36 @@ static void test_one_syscall(struct syscall_cfg *cfg) if (sve_vq != sve_vq_from_vl(sve_vl)) sve_vq = sve_vq_from_vl(sve_vl);

- ksft_test_result(do_test(cfg, sve_vl), + ksft_test_result(do_test(cfg, sve_vl, default_sme_vl, 0), "%s SVE VL %d\n", cfg->name, sve_vl); + + if (!(getauxval(AT_HWCAP2) & HWCAP2_SME)) + continue; + + for (sme_vq = SVE_VQ_MAX; sme_vq > 0; --sme_vq) { + sme_vl = prctl(PR_SME_SET_VL, sme_vq * 16); + if (sme_vl == -1) + ksft_exit_fail_msg("PR_SME_SET_VL failed: %s (%d)\n", + strerror(errno), errno); + + sme_vl &= PR_SME_VL_LEN_MASK; + + if (sme_vq != sve_vq_from_vl(sme_vl)) + sme_vq = sve_vq_from_vl(sme_vl); + + ksft_test_result(do_test(cfg, sve_vl, sme_vl, + SVCR_ZA_MASK | SVCR_SM_MASK), + "%s SVE VL %d/SME VL %d SM+ZA\n", + cfg->name, sve_vl, sme_vl); + ksft_test_result(do_test(cfg, sve_vl, sme_vl, + SVCR_SM_MASK), + "%s SVE VL %d/SME VL %d SM\n", + cfg->name, sve_vl, sme_vl); + ksft_test_result(do_test(cfg, sve_vl, sme_vl, + SVCR_ZA_MASK), + "%s SVE VL %d/SME VL %d ZA\n", + cfg->name, sve_vl, sme_vl); + } } }

@@ -299,14 +419,54 @@ int sve_count_vls(void) return vl_count; }

+int sme_count_vls(void) +{ + unsigned int vq; + int vl_count = 0; + int vl; + + if (!(getauxval(AT_HWCAP2) & HWCAP2_SME)) + return 0; + + /* Ensure we configure a SME VL, used to flag if SVCR is set */ + default_sme_vl = 16; + + /* + * Enumerate up to SVE_VQ_MAX vector lengths + */ + for (vq = SVE_VQ_MAX; vq > 0; --vq) { + vl = prctl(PR_SME_SET_VL, vq * 16); + if (vl == -1) + ksft_exit_fail_msg("PR_SME_SET_VL failed: %s (%d)\n", + strerror(errno), errno); + + vl &= PR_SME_VL_LEN_MASK; + + if (vq != sve_vq_from_vl(vl)) + vq = sve_vq_from_vl(vl); + + vl_count++; + } + + return vl_count; +} + int main(void) { int i; + int tests = 1; /* FPSIMD */

srandom(getpid());

ksft_print_header(); - ksft_set_plan(ARRAY_SIZE(syscalls) * (sve_count_vls() + 1)); + tests += sve_count_vls(); + tests += (sve_count_vls() * sme_count_vls()) * 3; + ksft_set_plan(ARRAY_SIZE(syscalls) * tests); + + if (getauxval(AT_HWCAP2) & HWCAP2_SME_FA64) + ksft_print_msg("SME with FA64\n"); + else if (getauxval(AT_HWCAP2) & HWCAP2_SME) + ksft_print_msg("SME without FA64\n");

for (i = 0; i < ARRAY_SIZE(syscalls); i++) test_one_syscall(&syscalls[i]); diff --git a/tools/testing/selftests/arm64/abi/syscall-abi.h b/tools/testing/selftests/arm64/abi/syscall-abi.h new file mode 100644 index 000000000000..bda5a87ad381 --- /dev/null +++ b/tools/testing/selftests/arm64/abi/syscall-abi.h @@ -0,0 +1,15 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +/* + * Copyright (C) 2021 ARM Limited. + */ + +#ifndef SYSCALL_ABI_H +#define SYSCALL_ABI_H + +#define SVCR_ZA_MASK 2 +#define SVCR_SM_MASK 1 + +#define SVCR_ZA_SHIFT 1 +#define SVCR_SM_SHIFT 0 + +#endif